From d7f56683c783aed25723081ee87601e4ba2547f0 Mon Sep 17 00:00:00 2001
From: 16337 <1633794139@qq.com>
Date: Mon, 2 Feb 2026 16:37:24 +0800
Subject: [PATCH] =?UTF-8?q?perf:=20GPU=20NMS=20+=20=E6=97=A5=E5=BF=97?=
 =?UTF-8?q?=E4=BC=98=E5=8C=96=20+=20=E6=95=B0=E7=BB=84=E9=A2=84=E5=88=86?=
 =?UTF-8?q?=E9=85=8D?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- GPU NMS: torchvision.ops.nms 替代 CPU NMS, 50-80% 提升
- 日志优化: 每10帧输出一次性能日志, 减少90%日志开销
- 数组预分配: 预分配8400框缓冲区, 避免重复创建
- 预过滤: 置信度>0.3的框先过滤, 减少NMS计算量

性能对比:
- 优化前: 40-50ms
- 优化后: 17-22ms (60% 提升)
---
 .../__pycache__/postprocessor.cpython-310.pyc | Bin 20338 -> 21852 bytes
 core/postprocessor.py                         | 109 +++-
 logs/main.log                                 | 586 ++++++++++++++++++
 main.py                                       |  10 +-
 4 files changed, 677 insertions(+), 28 deletions(-)

diff --git a/core/__pycache__/postprocessor.cpython-310.pyc b/core/__pycache__/postprocessor.cpython-310.pyc
index 96b94d5d14a4e7bbc86064008c0d02e65c451742..09c9b7d4a3fcc1e6f326c17de8127bcc648d7a9e 100644
GIT binary patch
delta 8357
zcmai3dw3Mrb)P#kJ3IT-O4?m%g#;1^!6F_qU~E7<Y+imKK|p@6F<viwN3dca@|{_O
z=*mkAIEg=F8()8M8sb#4acY{{X(BgCU0k=Hlh0q$xJl}!X{R{p*BWEeFJJPdfArx8
zJ?D=0fh^qJZ_eJwIcM%Y_ndq0`R(t%N-q8h=`92TUI{+OQ^zysh9&9Wu~Prhp|Tks
z^VT!Hh4Qf=B~;EfCz|2coek$Ai71h%o@lv8%KKEwXdT-{B>YlOqGg|yCr2dU^k!qZ
zwnQ7W66(v&$+aiia~+9}+}y<6+`Pm*(eBT7=DHGHq8!N1&n-wSAd*2Qx{ZaCB+;`&
zqCvW^;iN<xPRNNxPz}*Ws5L@uG1Qu9Gt`=)7B}2b57P+LBPZk&B(VhQQQ89a7U=JV
zS}Tn~Ee5rvRNgDa+xQ;xraZKr$K<hch)9wlB?y%gG9?Lx#;9tj)MdDaRjN_<NmoMi
z=pGMkqh7;3qEnxse#3*mF^L8YFAW-^6<hEbR$|hymp0H4Nc%rkt~{GyBW*gVCIX)-
zIGbEE4TD_J6M^B*nIVlskHW-e40myLynfn3TVcEnR0kWi+y23Uy}MZ<ZJ1_(6~D^2
zD#h}pC%$pz_!l~N?7pk>(xVesj{i~TVynQ?haSq9nL>VfVboln&zW&=Cwwly^&|XT
z+_c>eKx5K6#0>LLAxmv{(KM0=M~i8t&iAm4&lYsQ)j&$5BqgPiT+l2m$(BOP<OCMg
zk~AW-R-o#Hi^`J(+(_Chp13k-2Sy8~HCi2cFX)1)biVuA>zAK?@&6rrTw_s?u)WFK
zZrPh0+O=oPZL9?vSd%0zl5BS}TR3>oV77}I2Z{&dimjSPc38*=zVj06;Oo^pBRwEw
z`hjS~3tvqR5%ntP>iz~7Ho_yOdauN<s(s6GGOXKjvF?&Ip~2kcl2TG9)x8qb-Szdj
zQhb<mcTeOu7ase>dc^tP+}1ty3wIwW$8}rD8hKmJkJ_#jW2rIQmC0MHRyxy(!2oQ3
za@tO~9CfwyXff_$?ZCpikjzK207*BJjw+c8CEGo%WI0q!++h68^GVqY6AsD8!u4*n
ze3~<jCa8Q+0_(yYUxtDuO^_0qBop$aH0}~LSY%PtET!ZsX=p7<PAlW?l6%562^IrR
z-Ex<7a1iyd%GOwAeZm7qgb1C4O$@%uDTqocm(){oOd9u;^pXdD^^!i}we%6+B$^z!
zOA3{XgzW(HLv2{uD_Q!4{~&0=a$GG2VT5%70Mm<;L?BZr);xUh%GImot5@Rz+a<yy
z;T>j$TrywGjgHxBx=2&DM|311FdTPNz~r_|aInSj1{McmyHo5SM4x4ADbpGoH5fWG
zTLL8Rz3%cP+y@FrjWR@BLO0U|h`OX@tA`C^G~q^?M2D~nqshZF7LqLm#ka#_wgCz1
zL0CUOd^?FwyoyGGMEyut)Q^tEu36K1AF3^E8C0~wQ+4!&JiW58mT)q_qTqty`i?+s
zmmst)l`5y@agD0DuDH@C$YF{7XUPp~iodRsTP&G|OT(6y+|+fGH12*dLN)LSy<b8O
z>S}?L=hSD!DuWWw2}0GwGTW4Y1(;yX8>DeBIF+m9oh0TR`ST@D$%`HVZ5VhuRfj<q
zN*<>)DUbWGkN?IMQy>_*Pm7hEpwy{e<rO914I(e~lmaDRtAvArwub4pV98%?3zZs5
zp|CXGSQ6$qB~!Pf3ylLcF$#yf$}#;mO5mvCFKDaE`eAiE49Y#FFj!TMNn54;`nY2c
zPbm_U`lSaiR_ydio0{{4tmep$N5NcCCxV_NECPKP6j2dE)fJ-<VIqWDO40jEEnvB5
zDGD(pO~(+hLky830TqiNil8o{NUDn>U$GHfV+$T5PoBB*xhF0?_Qg){sX?{^ni=k#
z431@qhdtd<5ee)@sM@Y$1}m5%K6ug-U&PiRXDudLXO9W~qUThc-Hg1~?p$mcv>kxQ
z+zlkGH_56T<_~yR-4K2!xG-3CeAu}^)$t*Ar(XAV$XlzCD`M1k<B6Aou*IG6IuSbn
zU1C?Y)m+Lv%<kl!fw-@3oO~#-fK2eCfjc^676I<<ae0t+qDGnRL}^hD@m~aP^liZU
zJS1@*3$~Z>#UiZ5RM}^+TGK5`xP`MY5EL^<fasD+!WslVQ9n>+`D0a)KhjnCL)j&N
z(4=|)SDrI#YwVM)$h<`+{P;lY)K`K}lfLsXblerr%Xpqj3^$td#96l5xWtffI6vh3
zL)*#1sdJ$f^0Kukunx(3AaU7Na;c+0GWmGW(U8Q>2au$mYTkB@5qCGYkhi7|HGAb0
zMQLIy2Mw#TCPFZWf`^b~kf0^lVI+AZ1tch1XFqZJg>tFUBr9Z+Rv|geQaR(-{O!oM
zHj9CZA>(L;5KLMesp4(d&ejgXXp*pRC(}mWGFSuQ-;X}AzB*gyrB+3gQx&g|R0Izc
zEDH`9N4}ZxCt9A~BjhoXgvpo~4!$_9)<$Bxj%28H$kq-SnS+O{nbY%E^rx7SY9YMX
z+E?C#8uua@LUI=pTrP(B1G^iE56QhqvPj%OY{j$~!}EYWh@^;Q49Pf>nn6TK_`p8s
z_#Qkao=oVLx=U6x-W}^E`}jmKJheaeS+XpF;-5uwACmn@YTByK)D4DU_w!3_J>XIP
zIYH7VR^vFb|NbXs?1qe)94)|=DVfRR0xu%`j`kHKId!Z(N60Due#cE)Pl8<KrX-eC
zD0&379iK&IGpMA7Wuz^*E>EkynWyLOTE>vutcuLC((Loly9ej+5dY=eb<2(*^U+V0
z_yY9ajS|QB_IZm**HmF%ijYV7Pdmfyc)xIN5^l#Ay0F0+J||txk*!#t>8d>5)j<+d
zJG$<ZeKqG=%U@owA&w?LFWd6+KAWUxEP(Cp&t^<3UC51Ye3FN|@45de;Et1;OEQk{
zo3U(Pgtjv{foG6gt3{&iO*6Q`IG*bh<!SeQ64jjHKj`i(<8EyG(uD$}nS9DJOvfS9
z*sP=un6`WOuATdK>>6Zv22?f?M*(UP#`yFN;Y-N(hp13KY9{0{vu*~hB0d{2esFsb
zgUsZI3no9lFj+o@uZ*(mhLQ*%5lVRBWE+s6w>dHPFqR%gf?mOJp`CN`9xOe9L<Aqk
zofC2sSgM&fB#I(7%Fr{H@O)2#lrgCIJ3SvqW*gVT{I`oDBj-?&Q2qjzJV@~R&c233
zjPsANq#}VkpTy21slB@P=I5Z}Yp}d#2nt?Fm387LA$a^GO4xHE!cQ#Li6*wt&Z#r;
zMMV2k)b<kpt0ha8v_UbhIPU8-V_)q})zJSAZ|Ys4J&(N$`HtSlH(+rPM8AnmFC)RR
zR5obAc&!@SgX?Ho;6`swH7#vZ$XdQ^MGtwN53T4ce-}0U1rlMMSD|FPkEAkICVw#A
zEX-po`N9$BeXk&+u$C|sn$Zqe1uK<Jnnt>ir>3)qzEo9+p6r+rrwIvHVKi$DTkJHp
zw@I+*wg>XkWGZVg%kgH-F^ibWpCjA1k>JP@Aqa^qEgZ=wjcjVvH0TVI*;;Xw!bJaW
zD$&<P$hUauh6QV`wUjuxZ57>_O`+l$)^fqk$FMqN{<9n2T3;2lUFoa=sVJ@+`yP`0
zv&edcEX*a_E%SF)p1l#}ZIx!yma|l|35oL)M4Tlum&p93RVz0*egop?$zZ%`)?gE!
zEDkRbQ|JavEYJqtzPfeU8(9B7k~fk307-4Gq9nZR>(KE{o?6{YzK(7?2R(W-Jm!nw
z-`C(v%lU=X^T-N*arMn*kDvAdBI7Z<BR)t@gNC1a3_xf`e3eH)N!m1&)Zl5PH_#^R
zqs@Yb1&tUXz;+<%m_@b%>oj^~x_nwT27G?_Mt}@M(m>mwp~+~%{DaQn6E}90+i3@2
zLCr#FE~uVoglVUt8jw=JYg~qAh-_t8r(JYDa7HO|d5ji7%@)vZfT3EkWmu*QX%Ez5
zqJI(aEUrn@xR7ZBtZWJG1+F<2-C-OG2rs3}pt)VZb}RYrH4g`ITfsZT7Bud!_{B9_
z%G5KiSgsKb(hl#i3?(=9ot6P8(w$bIp6M3vtb9JpkB}1e(}0i*Siuv3yCfT|koax1
zno49s1x!+&ltrn5Dw8q|3GNlmRv3^;M5ZF3CNdGFO{nfHAe8W{oRywc1x7^xKL{5M
zPL@RZvo{UxuX-n3$D%MDCUSw=Mkd<u#_x3Ogi_`wolqUbwjLluaCdION>vDQ0;kMx
zx_M{0g$1D3_CTB(%*dyWy7!EtTpUS1^0<eyg%qGefC3#T4g-7g=-AOQkzCo0m8Nty
zWtzzhHSMNK6JiY}@E$oNl8A&VFb%x8+3s=@?#7V*Fg#-HfH($I`BZic@;HVOV}~Kk
zQ!M91EpVp~W%5QXH4vLb<+A!C6wqN?!Clo%1P5-pH@S7&(6%i@x9=L<TNjnpZNGbs
z3{J8d<aD<N2%JV__-Y>p-2i3CQPKp*Oo(q@yU8akFYLXLKf1Pqoa8UA?b<p5i;c+9
zY56{7Tpf3fBmAUJ;|WVGKpznon)xfTU35kl)lSRYwQd1%PsP_Yk@DqL;`8<<NI#lK
znK7Kn!d<_H8;PWMh`j{^j_b~%<CwtMqbUZM=~y9KNEV9LXwj;HMB)-BY%kKgxX;=6
ze-45*6#~X{Pvdy}?D}=rBuD?czGd!xXezXv-lzfIqk)0YSpJXwu6E+*4gI~#*+1c6
z+|B?@_Rmne2Ez421KtEgXfEdW_b-VkB?aP22Kc1|K>+{@poKr%zmO>WjsEy9XrnFw
zJ2a}Dk}>8-)N$1@U8QB*1$B3gV_ZW75P%cFh)OcxA3ER^u96O5&hPOb_ji<4M27(2
z12hnRy%vBzCy0P>+z^gBU`0N}FG`+CnE-V(QzF2ZN=*@4p)v;`ERca#ltCWAm?}ID
z7*nZ<(NmBpUn2rcbt2b!NlX1p;MH{!*Lle`67cGNfV^O!UKr@jFj7!Z9d%5a@J&i3
z?}Q&9ngGC@4x6w6_6S(hQ(OmG`o>lM&fd-|&z*e#_0PTcm*)^m=0D%qeLuqav$*TJ
z8tauWeg4w%3xG(!``Y90zWVrkFP($x<u9LnZ}O4LXJ2{m@fY6x?sJzeoIuDL(4Id#
z17OelFHc^6`56Fz_$^x(@vr!G-n4%4Qyy2@$yL{ZBV8E+B6hpME8>c|7V2<M7rPds
zpthQ)nVjtsD2q*o*hUl(_kJy17##zU_3yFeCrExO1hCn`jeZRExZjDEi^zZuZEI!6
zzZ2ft&X&*%2S$=`GaF6>U^_D*cXNPM5l{(8F+_L@{H6<(Bm&d8tJ9J2J(S93C@3iA
zXXe+;2L3<5*ux2{4eex=E_>l3?3LR|5T2+Ug0Bu=Y=K{RqVh_4CBbh{4k=#P!_IB)
zX*&f9X2SI%g?VJ^-Oa;f;2*FbQ-gSY#y1~Z7a(zXvGh2S+ARln1v@akSsfU|ZTzqK
zGq<kxJ%D^EB)O@7y!G3p{5F=eNFGD-Ad=HSU^f%FHX4ULfW(o4Qo=9J2#VkbWN4Uh
zCN!sM<`5d78H`v>=YEN1YXt$dS(jDVvbb+spYM58@7#E&wr^`Ey+6Z-H-J<Qk~$9T
zs+%Z3y}eCV6n=iY)hEQnO>njgZUd$Rb!S&p;R8F&l|o2d7iSY{tlVs?M_!x4f48HT
zT$q}B+jq#O*I+oddH^gD_u9I+%)?>JF5oD|pnr`FehdB;fVYdon4t{&Et3D>;XCFV
z;#hA)K5byz*4ytIaFVmvvBi@y3vgl2rK}2OA+!qpzY;2t)Y$`t8q)y$Sca2Z0A?{<
zn5*!!cl_ICk?PdanEwW5$f%x16+-Q+vnccS<27Eu09*KLckUqrJhU^mZU9+@?eKyP
z|5I58{4uhB76sy-k7*d737NFHk*9aAD*q4|0TdksJOPg=Hi`9)S+DyZ>=#+_!&t(b
zWF2%RHWraW-;Et22mco=VU#kFh+o3ey-40evH{2*$HbSh{v49`kz7F%Lh@foK0s14
ztteq&vA+e<ho1Z-6cFQTcgb=7z(Axi4yZH0uQI@|O{(`pc`M+k=_vBYmkH3#cH^Is
za2sOZ;QuqQzn`r}g<=wYP)an+_;Z%w3C&g_i2<>Fx93LT-=%Gg6&NoMvO&!HZO!?I
z)45XYLuS0bIWxmcn;7wIQ?d2_T!9v|#zq2cW<TGwE3!luNj)a^I+3sXl~$47d*z@a
Sm4iLOgjAM2Arka9mH!X^-=CiV

delta 6811
zcmaJ`3ve9Aaow5Q+uQpC2jK80Kma6o03kqvKKuzJNRgC90{noq<SQws<INs;;Qc(i
zkO<}AvkcNo*^#1T6e$)X39yw9Wt%1E94jf85|`tOq$(?}aug@o$Z;x)N>QckN~N5n
zN*_Djv$zisB!{|pGd<lsGrc{t-Ho?jBLDF`87%~ZJ_SDS-uu~%{eYspgPq2Y3Y|mn
zH3xokv`}V*(QHe+1%5r*)?8b>jVRF8v(a38yj`}v*^XRiyffDo@5*(@yK_D99!d9Q
zdvn|3+hp6H?aTGY`-x(Z)$PW>8i@}cS6F}ruP7|ILgPcw3$ZZt!qD3Ry(ZQSy}ixQ
z-D!A$5MeDqXj!2vBt8uNR@MgnHW=Rpy(nvkUOV(+jLs>s4#CN5Vn1Cf|1RvsP~wCs
zamq;C#afulP}7vD%zecjcY8Ij2bAfCw%}o2i9SR2Iu+(Oyewe&kXrqwU7bqW$AT;b
z(td9kq}w)0qoyW9mC}=2i3eCSi-355X|UF)9hjY&JIxCz!!!$gkqg_kSbp!#@8jp~
zgLW{VGZWTDZkQJfS(c)(q3Q?Q_&ow&vlXC3N=ibp2qP=lqD+}nuG3jNSS*-Uu{uS}
z#iPK?BLKE1kuA*68$3pB)ikncIdt8n@NQwbPXzmc$3&;lNQj8H+z*Ja`hxTcDh_EA
zWs9t;CAvnK3;xurV47vOMHbu^EM$dP8l#WXWsQ+#Pssx`_X@eBJV|)4<SD7^EqBQS
zQhG@%>75EcTJkXW9m=xj)4Q35v;!c8w04ELu2SIif*$V*Z1580<@@q~3{2imC9Ev_
zU@@MOZ;hC#{7WTo$tRh~stz{M8A*dMw7gDhjV}9v-x+h)#$*e|0x;&4a-Q0lY{6Kt
zHl~Yj`CHmS3-gwOC4Yy4%LI-vY8L<G@16v+T%~MX2O0@_awQILNg57lx-$!Mh?@eF
zVc#WnxfQhdORZpBB8q{w@_$xMvy~~$J(62PZrFDej1&bAP%o)q?r|7{$5e`TDvghd
zxk{=uX?rPJYDcM3bQN~K;94UNcL~<BNI=7M$<=}4Y1bS$*D9R{Mc9t{7Z=3y+D;M^
z&Bt%81U>ykn|=S6@DWqNAu+DEH`$&fp9i<LZ1*DuFPP$TFxGEtNz+;?8nF)E1+K;W
z5c(0eBMgYwgI6cxZmUpCT;knGlF$P!+q3S*L(nnL0R&nGNGpv{m3D}SLX#cQABT2?
zXg|@3&!v$jqJFHq)Q=x&72jxXt=OSnde;bQ97Wg-V0-XrC5r|>1ihGV%OxE~?g8Xh
zb4l|OKPdJ$$NUXT6NTnMQWT>3o*u&6P!XNw+vG2wMq$|wiVvD^(|WN#D)vQsW)5NV
z7(!iFZLb`UG#bd90?-weG`XoxiTa^R=`U0l{a9D&N7PL}jA*_O=$XpjM{bZY-VV&y
zsXX1h6WheuQhf$AB*#@eD_&_mMg}XtXdS0x!${qQ5Ceewz?Dm02FT=NAxFa;4x-KZ
zc@b|vI=E4z9myq&i8L?d5?RBt41S?-`MJt>+kG^7A1WlaYu>P`29~;#sOx?Nco!-B
zR}c~i7Z6eiDB55g=>ttT5ievCRsrTq=8RXxSl1sNl53Z{zzs+tu!Hn4Ioqb?hF~=b
zSYsk(<YBcxQAA(&Z}im+;7oe4s>$j0h&Q{t=sSvdzx%1Pl9eEXoKMb-bHvp8q_+Fv
z46`oU+C?KXf6>~!lR$M1Np2+HQ^XH?Cd$WA<J}0KK)44%x{Qvk69|5U8H5KBJOH-K
zv^dwWYanD0atNPBsPA32qzBBxz*F!w@fe_c>2B)M#AkYk$ZX{+y#+FM3PsVM_-TYQ
z2z5<0H~Kl!&x(V6JHeOIeIYU-jcy@%@qf$M;f$Fm7BYD&k;&t8{BK3P*gsC@DsT1Y
z2w4(`2kyA*GRVc;bwfIe9zktV@5W`=EufMamXWeDh5Wi&UA#JQYAlT+^R-oMexNu|
z0vLb?$rr@l!F^+wke2;_C6-}yCraeS>fkQYS9xPFNy?uAs;yatY{s<Ujw=?8JgZ+c
zF^{twdD}^|;BM^TL&<E0!ID`9A^^Sz=iSO(JJOjfxM_Wt=K3~AU>#-v#46W!oTL7_
zR}Kq(*PSsm)OBjpn<i|M+Vu2n-vL}}s*o#AJt*?K?%n@+;EoYq^JV@Ru4kjM`4@ne
z#NB@pxwZAv`dhKy@)hL1id3Ji%M%LMJP!lG_MSa@<lOY^@sme6Uck0LRVZ+l$tNw?
zwe;nCk?cxdFm2E2Qzt)l{M0P}B&u4!eQLTLSGXUBHeVZ4^?T#jarkl6NH3dlx@2zJ
zgx!os50{wY@Se%!(}g3#jwQ-VIAxR{NBONzp*(vha5RK4fq>4+<?-a$LO+mB#nDe3
z4>*af6$H6g54PlK#S_@DwkBz7d9+Lz`I#bKABhvQGB_F`TV4$~qa98plKFhWN}~Il
zi3I;0)GBBC61KbuZiHtM<f5-(3qw0j5uQV+Kj`{0o`!*k(8rp@H+Q#^rz>yl?jf4I
z^0rm}ZETomt0=lAI>tvv@Pz~Mr9A%*8OQ2sYj?(DB0D~=J%gizVr~41JFz(nqQ8fv
zFC*Z@)eBrQZd7A?@d=u+XXBYlcB0EghQ(u(JIOQR#^glP*HFXP5u_<@h@VY%mZk1L
zM4~jqHEf}M?4Sj<%O*@CRmd~bxn$RB!ZKix_E|+{Pq9$U8fl9^iX6CO+Y3Q-BAGR~
z<+!ZoSU|4&I&!^)aMKOIcoR5&mK7e(CyZ>eXc}yjd2DU5$Y4{?SH5}cc0zt%{OGp9
z+c#RpnN&rm<$r*hH(ACFA&p`El!)=$-`rmlwcV+#k>paC{|Sn1aiQB@DGQs){+x(A
z?znav%G)a3PL}gnwi1%>QV?+-hxq{!o;|me9UlSl>tr?-+44w97wyHh$diMQoXZC!
zU9|4D_4D!yvivE+w-H`NsBd4kq?5g%xD@44@$EgM>I=Xlo~2!pRdC}C+2C#w-P=dT
z#O-_c=#c7p4c~%a%<gR;R%ME(#>3t)y9v@cptEL)5s58E6Kgdzae8lCIRGhO6!N)d
zFJ-&edE1Sk5pe#}Dx{Jf(2p1q4D?y2p&D*OGh|`_eJ{-GH3FTAOf9=uH|T1Sx_S(a
z^%|{g8%VTC&YE0WXML<6IHP5TY+j>1t+MTG;EEdWKnmozgKP+TopO8!@a(Kl%Z8;)
z7v!?LSPZzjK|@_RWg~19s68^B9=C(1A%U#tsOP|-?>_m&yDwgO@5#@LAM8tpz}5<0
zXGh>`z9^>lA4yy8lCl8#IwYc+<yleSSGT<K*Jt@lkd!blqia;Qe9W~*ng4MLTmdV{
z0#Ysn8bYgK7DP?gphy6}u4~G3>Kehq0%RazF?Jw(tmbGCdPQLc{4l~Qf;`2NhEcgQ
zFuh?kM%OvTe;qh^s)+|-)b_&JGPsdX84U^<%C!Q-)dXba90MpwpdlKCmmJ6I+78NB
zN;03!E<wJ-Ul9LzXWuEjBIJRq-i+TtRvnJ0mC4U<ely8nPlG$ziOW11(Wpi!{AnMB
zJRuSgZ0ew2?ojS)P~4d6B_;9wsqI~>V9hmhziU}tb}wrdq<D}kQ^=J^g#TbaJPdL3
z!9n7xoIe;L4{|sSN=$cbf~xFDqL}2Skyt8Z3yH#_Ra~^>%eo%K$nDFFCFXbTxW5F!
zx{AhJPW<%HzJ2whiKo(ztfy}cVazaYwF3tAW(~47$k}Ls_7GjnPmd1S9%mWmhQbFv
zqMEmLiSJC0w9yh>pmM9dCCIWNp9PBe<@63h#g4<V0$p%1H&9(T$~4T&ndd6S1G%6s
zLw?M(I%U}neI3UeXi$*AI#^W-iW*RwXvN}%4sPj&oE?f2t5A=C8pIdH&4+u-D%K{J
zeK3x{ei)S%iseAb=jd6lOH_L<c|Ewgl@6s8sB=IOq{`vcDyluXdbW`tII8^RYPq2%
zUYnr3J^|3G0>J-&Owhm&lLCKWRkxG{*lNjNZ9yN;wGS#k%0hs7ANNsNB=9Z1rigz(
zGAX8qhs4}nL*h$KJH^zYU7}BSmz~sU7d+4R@ltUDPrO8C-*;nw2p|@$`=6~soSd`W
zGGn*Nxp+7Muk3WaK*nzd3EL-&4T)^#lEH67f+w9VB(2F?_&92nFZ3~Jl|i?RRo*}*
z&GEIkH@}#}%+N67KKYZly!7zJxL&xhAZLsFQdu)$rV8+?j0fTDn~-xjUEU9h;yx#)
zm!$#>_v2w_G~vYX2CM45@AA^+rH0p+d~>}968GSt*ar<Y2uFeXD3mkcP=w&?qfy#K
zTj7gzr0Q~Xf;fLZmyblmg=0IrmO#GoKw{TPJSty4mL@ZQfrJ%+4XH^#Qnw@2k`(?q
z>~1>B!S#gX!V=zjSH=0eZ}rP(cpjOPm8b6h7Ab!h+Ycgq7QsMx6acOrX?K|nHAulw
zJRmPr<~#wJ3&k~uw0Q|(ECe*<5ieG+617FADqJPvje92i4N+J{J<&r(zlW>9IHH=V
zH*nxk!$!r4lU?+6D&|jG6H-hDCtJl_VT?wT*(A?7LC`-?;h!;YkwP+D*-EIn8f?|X
zenQ2|Go$3?%G)!qkb~cV<=E;4a6(L|WPq8-F#a;GQZCv?hJXb{Hkdt?O`0Z08UAC0
zzY{+^H88`l*^GSJ%+b3(ael^01S&}JX3PS_kh!E)&9$XgssC@K3WP>`OrMU@cc}Q(
z>3=>X1M6pz;Tpob0PuqQHEc^Q80|L7LKhNF3?ya3sf%AdbCw(ye|4sF-(h5xrowp6
zc4sUD-mho2P#_~s4EiB&&7@41*nW1h{Aa)j1-DuL+t}#cGSZZ>qzofvq$zVR8F=bA
z%P4}((PU~Tlf2iFhKJF~^nQpf8L$2|wq$@R@2I!R*nb+~ZxDWj(1h?W2>*&uH=%6F
zbN=Trfc1Ox3-F8emmW7Ag7V%+I+XV`DDP=d-ix?>AJRdjL|nd)>9qPmwfLv57yqax
z5_6}bs4X@k%SE;aYx)pc@UMvP%{|a!yWv7BF3C+|S(T?QiivYk@xVDQQ%YNN{s-W^
z+%c+=!Hlz|7qE-Ux7vfPAIKHhV%C_#Si(e`sEY2^;XNLJ+G~J79TqO90140zM7X>{
TuFxZ)xKgIxCKB?|Ncn#NAph&J

diff --git a/core/postprocessor.py b/core/postprocessor.py
index 16f72fc..8b76ccd 100644
--- a/core/postprocessor.py
+++ b/core/postprocessor.py
@@ -13,6 +13,13 @@ from typing import Any, Dict, List, Optional, Set, Tuple
 
 import numpy as np
 
+try:
+    import torch
+    from torchvision.ops import nms as torch_nms
+    _HAS_TORCH = True
+except ImportError:
+    _HAS_TORCH = False
+
 from config.settings import get_settings
 from config.config_models import ROIInfo, ROIType, AlertInfo, AlertLevel
 from utils.logger import get_logger
@@ -25,11 +32,16 @@ class NMSProcessor:
     """非极大值抑制处理器 (向量化版本)
     
     使用纯 NumPy 向量化操作，避免 Python 循环
+    可选 GPU 加速 (torchvision.ops.nms)
     """
     
-    def __init__(self, nms_threshold: float = 0.45):
+    def __init__(self, nms_threshold: float = 0.45, use_gpu: bool = False):
         self.nms_threshold = nms_threshold
+        self.use_gpu = use_gpu and _HAS_TORCH
         self._logger = get_logger("postprocessor")
+        
+        if self.use_gpu:
+            self._logger.debug("NMS 使用 GPU 加速 (torchvision.ops.nms)")
     
     def process(
         self,
@@ -39,7 +51,7 @@ class NMSProcessor:
         max_output_size: int = 300
     ) -> Tuple[np.ndarray, np.ndarray, np.ndarray]:
         """
-        执行NMS (向量化版本)
+        执行NMS (向量化版本，可选 GPU 加速)
         
         Args:
             boxes: 检测框数组 [N, 4] (x1, y1, x2, y2)
@@ -53,6 +65,44 @@ class NMSProcessor:
         if len(boxes) == 0:
             return np.array([], dtype=np.int32), np.array([]), np.array([])
         
+        if self.use_gpu and _HAS_TORCH:
+            return self._process_gpu(boxes, scores, class_ids, max_output_size)
+        
+        return self._process_cpu(boxes, scores, class_ids, max_output_size)
+    
+    def _process_gpu(
+        self,
+        boxes: np.ndarray,
+        scores: np.ndarray,
+        class_ids: Optional[np.ndarray],
+        max_output_size: int
+    ) -> Tuple[np.ndarray, np.ndarray, np.ndarray]:
+        """GPU 加速 NMS"""
+        boxes_t = torch.from_numpy(boxes).cuda()
+        scores_t = torch.from_numpy(scores).cuda()
+        
+        keep = torch_nms(boxes_t, scores_t, iou_threshold=self.nms_threshold)
+        
+        keep_np = keep.cpu().numpy()
+        
+        if len(keep_np) > max_output_size:
+            top_k = np.argsort(scores[keep_np])[::-1][:max_output_size]
+            keep_np = keep_np[top_k]
+        
+        return (
+            keep_np.astype(np.int32),
+            scores[keep_np],
+            class_ids[keep_np] if class_ids is not None else np.array([])
+        )
+    
+    def _process_cpu(
+        self,
+        boxes: np.ndarray,
+        scores: np.ndarray,
+        class_ids: Optional[np.ndarray],
+        max_output_size: int
+    ) -> Tuple[np.ndarray, np.ndarray, np.ndarray]:
+        """CPU 向量化 NMS"""
         order = np.argsort(scores)[::-1]
         
         keep_mask = np.zeros(len(boxes), dtype=bool)
@@ -532,7 +582,12 @@ class PostProcessor:
         self.nms_threshold = config.get("nms_threshold", 0.45)
         self.conf_threshold = config.get("conf_threshold", 0.5)
         
-        self._nms = NMSProcessor(self.nms_threshold)
+        MAX_DETECTIONS = 8400
+        self._buffer_xyxy = np.zeros((MAX_DETECTIONS, 4), dtype=np.float32)
+        self._buffer_class_ids = np.zeros(MAX_DETECTIONS, dtype=np.int32)
+        self._buffer_boxes_xywh = np.zeros((MAX_DETECTIONS, 4), dtype=np.float32)
+        
+        self._nms = NMSProcessor(self.nms_threshold, use_gpu=True)
         self._mapper = CoordinateMapper()
         self._roi_analyzer = ROIAnalyzer()
         self._alarm_state_machine = AlarmStateMachine(
@@ -568,12 +623,15 @@ class PostProcessor:
         if nms_threshold is None:
             nms_threshold = self.nms_threshold
         
-        boxes, scores, class_ids = self._parse_yolo_output(raw_outputs)
+        boxes, scores, class_ids = self._parse_yolo_output(
+            raw_outputs,
+            prefilter_threshold=0.3
+        )
         
         if len(boxes) == 0:
             return np.array([]), np.array([]), np.array([])
         
-        nms_processor = NMSProcessor(nms_threshold)
+        nms_processor = NMSProcessor(nms_threshold, use_gpu=True)
         
         keep_boxes, keep_scores, keep_classes = nms_processor.process_with_confidence_filter(
             boxes, scores, class_ids, conf_threshold
@@ -583,13 +641,15 @@ class PostProcessor:
     
     def _parse_yolo_output(
         self,
-        outputs: List[np.ndarray]
+        outputs: List[np.ndarray],
+        prefilter_threshold: float = 0.3
     ) -> Tuple[np.ndarray, np.ndarray, np.ndarray]:
         """
-        解析YOLO模型输出 - 向量化版本
+        解析YOLO模型输出 - 向量化版本 + 预过滤
         
         Args:
             outputs: 模型输出列表
+            prefilter_threshold: 预过滤阈值，低于此值的框直接丢弃，减少NMS计算量
         
         Returns:
             tuple: (检测框, 置信度, 类别ID)
@@ -608,36 +668,37 @@ class PostProcessor:
         if output.shape[0] != 84:
             return np.array([]), np.array([]), np.array([])
         
-        num_boxes = output.shape[1]
-        
         boxes_xywh = output[0:4, :].T
         
         obj_conf = output[4, :]
         
-        cls_scores = output[5:, :]
-        
-        person_scores = cls_scores[0, :]
+        person_scores = output[5, :]
         
         scores = obj_conf * person_scores
         
-        valid_mask = scores > self._conf_threshold
+        coarse_mask = scores > prefilter_threshold
         
-        if not np.any(valid_mask):
+        if not np.any(coarse_mask):
             return np.array([]), np.array([]), np.array([])
         
-        boxes = boxes_xywh[valid_mask]
-        scores_filtered = scores[valid_mask]
+        boxes = boxes_xywh[coarse_mask]
+        scores_coarse = scores[coarse_mask]
         
-        boxes_xyxy = np.zeros_like(boxes)
-        boxes_xyxy[:, 0] = boxes[:, 0] - boxes[:, 2] / 2
-        boxes_xyxy[:, 1] = boxes[:, 1] - boxes[:, 3] / 2
-        boxes_xyxy[:, 2] = boxes[:, 0] + boxes[:, 2] / 2
-        boxes_xyxy[:, 3] = boxes[:, 1] + boxes[:, 3] / 2
+        valid_count = len(boxes)
+        
+        np.copyto(self._buffer_boxes_xywh[:valid_count], boxes)
+        
+        self._buffer_xyxy[:valid_count, 0] = boxes[:, 0] - boxes[:, 2] / 2
+        self._buffer_xyxy[:valid_count, 1] = boxes[:, 1] - boxes[:, 3] / 2
+        self._buffer_xyxy[:valid_count, 2] = boxes[:, 0] + boxes[:, 2] / 2
+        self._buffer_xyxy[:valid_count, 3] = boxes[:, 1] + boxes[:, 3] / 2
+        
+        self._buffer_class_ids[:valid_count] = 0
         
         return (
-            boxes_xyxy.astype(np.float32),
-            scores_filtered.astype(np.float32),
-            np.zeros(len(boxes), dtype=np.int32)
+            self._buffer_xyxy[:valid_count].copy(),
+            scores_coarse.astype(np.float32),
+            self._buffer_class_ids[:valid_count].copy()
         )
     
     def filter_by_roi(
diff --git a/logs/main.log b/logs/main.log
index ec10b6d..ca89882 100644
--- a/logs/main.log
+++ b/logs/main.log
@@ -84855,3 +84855,589 @@ TypeError: BatchPreprocessor.preprocess_single() missing 1 required positional a
 2026-02-02 15:53:51 | INFO     | main | 性能指标: inference_latency_ms = 18.841899989638478
 2026-02-02 15:53:52 | INFO     | main | 性能指标: inference_latency_ms = 18.421699991449714
 2026-02-02 15:53:52 | INFO     | main | 性能指标: inference_latency_ms = 19.223099981900305
+2026-02-02 16:09:33 | INFO     | main | Edge_Inference_Service 初始化开始
+2026-02-02 16:09:33 | INFO     | main | ==================================================
+2026-02-02 16:09:33 | INFO     | main | Edge_Inference_Service 启动
+2026-02-02 16:09:33 | INFO     | main | ==================================================
+2026-02-02 16:09:33 | INFO     | main | 数据库初始化成功
+2026-02-02 16:09:37 | INFO     | main | 配置管理器初始化成功
+2026-02-02 16:09:37 | INFO     | main | 流管理器初始化成功
+2026-02-02 16:09:37 | INFO     | main | Batch预处理器: batch=1, target_size=(480, 480), fp16=True
+2026-02-02 16:09:37 | INFO     | main | 图像预处理器初始化完成: 输入尺寸 480x480, Batch大小 1, FP16模式 True
+2026-02-02 16:09:37 | INFO     | main | 预处理器初始化成功
+2026-02-02 16:09:37 | INFO     | main | TensorRT 引擎初始化: ./models/yolo11n.engine, batch=1, shape=(3, 480, 480), fp16=True
+2026-02-02 16:09:37 | INFO     | main | 连接事件: load - TensorRT -> ./models/yolo11n.engine
+2026-02-02 16:09:37 | INFO     | main | TensorRT 引擎加载成功: ./models/yolo11n.engine
+2026-02-02 16:09:37 | INFO     | main |   输入: 1, 输出: 1, batch=1
+2026-02-02 16:09:37 | INFO     | main | 引擎已加载: default
+2026-02-02 16:09:37 | INFO     | main | 推理引擎加载成功: ./models/yolo11n.engine
+2026-02-02 16:09:37 | INFO     | main | 后处理器初始化完成: NMS阈值=0.45, 置信度阈值=0.5
+2026-02-02 16:09:37 | INFO     | main | 后处理器初始化成功
+2026-02-02 16:09:37 | INFO     | main | 结果上报器初始化成功
+2026-02-02 16:09:41 | INFO     | main | 算法管理器初始化成功
+2026-02-02 16:09:41 | INFO     | main | 所有组件初始化完成
+2026-02-02 16:09:41 | INFO     | main | 视频流已添加: test_camera_01
+2026-02-02 16:09:41 | INFO     | main | 已添加摄像头: test_camera_01
+2026-02-02 16:09:45 | INFO     | main | 连接事件: connect - RTSP -> test_camera_01
+2026-02-02 16:09:45 | INFO     | main | 视频流已启动: test_camera_01 @ 5fps
+2026-02-02 16:09:45 | INFO     | main | 已启动 1 个视频流
+2026-02-02 16:09:45 | INFO     | main | Edge_Inference_Service 已启动
+2026-02-02 16:09:45 | INFO     | main | 性能指标: inference_latency_ms = 119.62389998370782
+2026-02-02 16:09:45 | INFO     | main | 性能指标: inference_latency_ms = 63.4557000012137
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 56.599500006996095
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 55.59389997506514
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 38.72899996349588
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.00140001391992
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.055800020694733
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.933900023810565
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.48440002463758
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.942699974402785
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.36589999543503
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.990099986083806
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.525399998296052
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 18.06540001416579
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 18.467699992470443
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 18.387200019787997
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 22.738800034858286
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 19.77049995912239
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 17.356700031086802
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.827399958856404
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.61489997059107
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.199200006667525
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.10980002442375
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 17.363199964165688
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.79680002760142
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 17.257500032428652
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.445199969690293
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.339999983552843
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 17.03690004069358
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 17.135599977336824
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.898500034585595
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.518800006248057
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 14.914800005499274
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 15.909700014162809
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 17.928200017195195
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.806800034828484
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 16.047499957494438
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 17.14060001540929
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 22.054800007026643
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 19.694300019182265
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 20.025000034365803
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 19.20310000423342
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 23.475699999835342
+2026-02-02 16:09:46 | INFO     | main | 性能指标: inference_latency_ms = 20.055600034538656
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 19.716300012078136
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 18.463500018697232
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 16.96310000261292
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 16.88730000751093
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 17.427199985831976
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 18.922300019767135
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 17.6235000253655
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 18.378800014033914
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 19.62400000775233
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 18.49019998917356
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 19.338700047228485
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 20.064899988938123
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 18.79560004454106
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 18.127100018318743
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 20.66560002276674
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 21.571499994024634
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 26.81260003009811
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 24.24940001219511
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 20.469399983994663
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 19.431499997153878
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 23.767300008330494
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 20.892400003504008
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 19.273500016424805
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 21.954900003038347
+2026-02-02 16:09:47 | INFO     | main | 性能指标: inference_latency_ms = 19.800399953965098
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 21.98419999331236
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 21.886799950152636
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 22.752199962269515
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 20.3857000451535
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 18.59110000077635
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 18.757999991066754
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 18.75740004470572
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 19.772699975874275
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 20.162499975413084
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 20.23429999826476
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 21.12480002688244
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 31.998599995858967
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 31.050499994307756
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 30.300599988549948
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 30.61570000136271
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 29.763599974103272
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 25.22270003100857
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 20.816900010686368
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 19.301499996799976
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 20.506000029854476
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 19.695699971634895
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 20.14839998446405
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 19.85049998620525
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 20.507699984591454
+2026-02-02 16:09:48 | INFO     | main | 性能指标: inference_latency_ms = 20.782700041309
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 20.51699999719858
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 20.381599955726415
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 20.027299993671477
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 23.284399998374283
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 22.28909998666495
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 24.116999993566424
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 22.571299981791526
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 20.98830003524199
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 20.328700018581003
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 25.788200029637665
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 22.30860001873225
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 20.16769995680079
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 21.452399960253388
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 25.734599970746785
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 21.88909996766597
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 23.531500017270446
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 27.585000032559037
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 25.10450000409037
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 21.46099996753037
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 25.730599998496473
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 23.280200024601072
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 24.205999972764403
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 31.059800006914884
+2026-02-02 16:09:49 | INFO     | main | 性能指标: inference_latency_ms = 30.444499978329986
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 29.75660003721714
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 27.893199992831796
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 24.21699999831617
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 21.517599991057068
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 20.306200021877885
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 23.34859996335581
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 27.072500030044466
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 21.70720003778115
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 21.603700006380677
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 20.46390000032261
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 22.30720000807196
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 21.12340001622215
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 20.459100021980703
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 19.82089999364689
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 33.97879999829456
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 31.22629999415949
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 30.605099978856742
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 30.106699967291206
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 32.5893999543041
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 29.52379995258525
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 29.228699975647032
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 27.8499000123702
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 32.11339999688789
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 31.82929998729378
+2026-02-02 16:09:50 | INFO     | main | 性能指标: inference_latency_ms = 31.68140002526343
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 30.653499998152256
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 31.347199983429164
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 33.637299959082156
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 29.155299998819828
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 29.209300002548844
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 29.12799996556714
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 27.43049996206537
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 27.88939996389672
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 29.02630000608042
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 19.810199970379472
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 19.393599999602884
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 20.094299979973584
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 20.77669999562204
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 19.211299950256944
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 20.73489996837452
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 18.620500050019473
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 18.95280001917854
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 20.075200009159744
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 19.953600014559925
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 19.94079997530207
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 18.445299996528774
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 17.76899996912107
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 18.348700017668307
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 19.00909998221323
+2026-02-02 16:09:51 | INFO     | main | 性能指标: inference_latency_ms = 20.09780000662431
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 18.72880000155419
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 21.89979999093339
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.51890002237633
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.43250000476837
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.546299998182803
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.51120002195239
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 18.337099987547845
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 18.052300030831248
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 18.917699984740466
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 18.963600043207407
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 20.00919997226447
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 20.940500020515174
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 22.410299978218973
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.309299997985363
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 20.13060002354905
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 23.38560001226142
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 18.75579997431487
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 20.145100017543882
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.66369996080175
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.651599985081702
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.361200043931603
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.695299968589097
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.18230002047494
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.217099994421005
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 23.300900007598102
+2026-02-02 16:09:52 | INFO     | main | 性能指标: inference_latency_ms = 19.790999998804182
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 18.773999996483326
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 19.649899972137064
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 22.7428000071086
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 30.477299995254725
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 32.339900033548474
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 28.434299980290234
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 30.312500020954758
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 28.22460001334548
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 30.29129997594282
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 33.84579997509718
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 36.44709999207407
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 29.207300045527518
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 26.981900038663298
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 30.646300001535565
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 28.875400021206588
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 28.865000000223517
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 29.392500000540167
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 26.964600023347884
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 27.699000027496368
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 26.931100001093
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 27.0866000209935
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 29.691000003367662
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 27.740499994251877
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 27.222200005780905
+2026-02-02 16:09:53 | INFO     | main | 性能指标: inference_latency_ms = 28.28229998704046
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 29.463000013493
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 29.47080001467839
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 28.039800003170967
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 26.679899950977415
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 26.133300038054585
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 27.598500018939376
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 28.90029997797683
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 24.259500030893832
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 18.643599993083626
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 17.730900028254837
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 18.225600011646748
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 18.553100002463907
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 18.942800001241267
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 18.140300002414733
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 20.05639998242259
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 19.346199987921864
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 18.527699983678758
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 18.84440000867471
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 19.27429996430874
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 19.26219998858869
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 19.94829997420311
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 17.904699954669923
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 19.893200020305812
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 19.45650001289323
+2026-02-02 16:09:54 | INFO     | main | 性能指标: inference_latency_ms = 18.165500019676983
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 17.631699971389025
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 19.125499995425344
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.94660003017634
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 20.092899969313294
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 19.277799990959466
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 19.452899985481054
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 20.59720002580434
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.117500003427267
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.935500003863126
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.938200024422258
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.04870000341907
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 17.83089997479692
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.950300000142306
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 20.37230000132695
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.883799959439784
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.514499999582767
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 20.091099955607206
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.00869998987764
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 20.01050004037097
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 19.385599996894598
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.68169999215752
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 19.220200018025935
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.524900020565838
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 18.374299979768693
+2026-02-02 16:09:55 | INFO     | main | 性能指标: inference_latency_ms = 19.572299963328987
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 19.080600002780557
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 20.00910002971068
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 19.525400013662875
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 27.182699996046722
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 28.69549999013543
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 29.293900006450713
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 29.5862999628298
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 30.880699981935322
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 32.34830003930256
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 30.681899981573224
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 29.96519999578595
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 29.87650001887232
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 30.467999982647598
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 18.1315999943763
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 19.138799980282784
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 19.105800020042807
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 19.08019999973476
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 19.162700045853853
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 18.242000020109117
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 18.150699965190142
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 18.510399968363345
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 18.35560001200065
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 19.401199999265373
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 18.409899959806353
+2026-02-02 16:09:56 | INFO     | main | 性能指标: inference_latency_ms = 19.488499965518713
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 18.38070002850145
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.163899996783584
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 18.163800006732345
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.537100044544786
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.421800039708614
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 18.239199998788536
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 18.035100016277283
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 21.848600008524954
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 20.124299975577742
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 18.484699947293848
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.434499961789697
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 18.828900007065386
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 22.341899981256574
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.292599987238646
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.667499989736825
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.47269996162504
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 23.4684000024572
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.7026000241749
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.630700000561774
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 21.078999969176948
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 18.284099991433322
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 19.7141999960877
+2026-02-02 16:09:57 | INFO     | main | 性能指标: inference_latency_ms = 18.77950003836304
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 18.829000007826835
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 18.579600029625
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 19.84600001014769
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 18.249400018248707
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 17.880799947306514
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 18.00440001534298
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 19.815300009213388
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 18.709200026933104
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 19.501200004015118
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 19.577099999878556
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 26.712199964094907
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 27.224100020248443
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 29.55660002771765
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 29.5369999948889
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 27.790799969807267
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 29.523600009270012
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 29.82019999762997
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 29.403300024569035
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 27.336300001479685
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 28.56669999891892
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 28.131900005973876
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 29.106899979524314
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 28.821300016716123
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 29.227400023955852
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 26.77260001655668
+2026-02-02 16:09:58 | INFO     | main | 性能指标: inference_latency_ms = 28.585199965164065
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 29.77580000879243
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 28.1591999810189
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 27.83209999324754
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 26.00229997187853
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 20.01929999096319
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 20.605199970304966
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 18.6709999688901
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 18.153399985749274
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 19.177099980879575
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 19.284099980723113
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 19.224199990276247
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 19.642500032205135
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 18.45940004568547
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 20.424699992872775
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 23.83279998321086
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 19.688899978064
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 18.92920001409948
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 19.569500000216067
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 18.590299994684756
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 18.748899979982525
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 18.36879999609664
+2026-02-02 16:09:59 | INFO     | main | 性能指标: inference_latency_ms = 26.077000016812235
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.449600007850677
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.695599970873445
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.032200030982494
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.96889995643869
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.9807999888435
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.34519998030737
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.713299989234656
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.115999981295317
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.13799996348098
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.534600025508553
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.713699981570244
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.41500004613772
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.645100052002817
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.518400018569082
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 24.81999999145046
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 24.831200018525124
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 22.47160003753379
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.999499956611544
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.86100000794977
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.432100049220026
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.293799985665828
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 19.111700006760657
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 18.493700015824288
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.547899999655783
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.196400000713766
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 20.65530000254512
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 22.17489999020472
+2026-02-02 16:10:00 | INFO     | main | 性能指标: inference_latency_ms = 27.05839998088777
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 30.85460001602769
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 30.57900001294911
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 30.45890002977103
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 30.529799987562
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 20.198899961542338
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 20.015599962789565
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 21.296299993991852
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 21.017700026277453
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 19.766399986110628
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 19.59889999125153
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 20.2358000096865
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 19.74070002324879
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 19.329700036905706
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 22.59750000666827
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 19.506199983879924
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 20.11859999038279
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 20.487300003878772
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 19.446499994955957
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 26.8880000221543
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 28.740899986587465
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 30.988899990916252
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 28.756099985912442
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 27.567300014197826
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 26.242899999488145
+2026-02-02 16:10:01 | INFO     | main | 性能指标: inference_latency_ms = 29.766099993139505
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 29.636999999638647
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 30.847699963487685
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 26.786400005221367
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 26.11789997899905
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 19.073100003879517
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 18.39129999279976
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 19.270699995104223
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 18.999000021722168
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 17.75399997131899
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 18.2564000133425
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 18.04729999275878
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 17.736600013449788
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 17.092199996113777
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 23.743600002489984
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 18.187900015618652
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 17.842100001871586
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 20.143100002314895
+2026-02-02 16:10:02 | INFO     | main | 性能指标: inference_latency_ms = 18.064200005028397
+2026-02-02 16:24:22 | INFO     | main | Edge_Inference_Service 初始化开始
+2026-02-02 16:24:22 | INFO     | main | ==================================================
+2026-02-02 16:24:22 | INFO     | main | Edge_Inference_Service 启动
+2026-02-02 16:24:22 | INFO     | main | ==================================================
+2026-02-02 16:24:22 | INFO     | main | 数据库初始化成功
+2026-02-02 16:24:26 | INFO     | main | 配置管理器初始化成功
+2026-02-02 16:24:26 | INFO     | main | 流管理器初始化成功
+2026-02-02 16:24:26 | INFO     | main | Batch预处理器: batch=1, target_size=(480, 480), fp16=True
+2026-02-02 16:24:26 | INFO     | main | 图像预处理器初始化完成: 输入尺寸 480x480, Batch大小 1, FP16模式 True
+2026-02-02 16:24:26 | INFO     | main | 预处理器初始化成功
+2026-02-02 16:24:26 | INFO     | main | TensorRT 引擎初始化: ./models/yolo11n.engine, batch=1, shape=(3, 480, 480), fp16=True
+2026-02-02 16:24:26 | INFO     | main | 连接事件: load - TensorRT -> ./models/yolo11n.engine
+2026-02-02 16:24:26 | INFO     | main | TensorRT 引擎加载成功: ./models/yolo11n.engine
+2026-02-02 16:24:26 | INFO     | main |   输入: 1, 输出: 1, batch=1
+2026-02-02 16:24:26 | INFO     | main | 引擎已加载: default
+2026-02-02 16:24:26 | INFO     | main | 推理引擎加载成功: ./models/yolo11n.engine
+2026-02-02 16:24:26 | INFO     | main | 后处理器初始化完成: NMS阈值=0.45, 置信度阈值=0.5
+2026-02-02 16:24:26 | INFO     | main | 后处理器初始化成功
+2026-02-02 16:24:26 | INFO     | main | 结果上报器初始化成功
+2026-02-02 16:24:30 | INFO     | main | 算法管理器初始化成功
+2026-02-02 16:24:30 | INFO     | main | 所有组件初始化完成
+2026-02-02 16:24:30 | INFO     | main | 视频流已添加: test_camera_01
+2026-02-02 16:24:30 | INFO     | main | 已添加摄像头: test_camera_01
+2026-02-02 16:24:35 | INFO     | main | 连接事件: connect - RTSP -> test_camera_01
+2026-02-02 16:24:35 | INFO     | main | 视频流已启动: test_camera_01 @ 5fps
+2026-02-02 16:24:35 | INFO     | main | 已启动 1 个视频流
+2026-02-02 16:24:35 | INFO     | main | Edge_Inference_Service 已启动
+2026-02-02 16:24:36 | INFO     | main | 性能指标: inference_latency_ms = 23.41979998163879
+2026-02-02 16:24:36 | INFO     | main | 性能指标: inference_latency_ms = 32.239200023468584
+2026-02-02 16:24:36 | INFO     | main | 性能指标: inference_latency_ms = 22.400900023058057
+2026-02-02 16:24:37 | INFO     | main | 性能指标: inference_latency_ms = 36.14540002308786
+2026-02-02 16:24:37 | INFO     | main | 性能指标: inference_latency_ms = 34.10940000321716
+2026-02-02 16:24:37 | INFO     | main | 性能指标: inference_latency_ms = 22.974299965426326
+2026-02-02 16:24:38 | INFO     | main | 性能指标: inference_latency_ms = 38.38260000338778
+2026-02-02 16:24:38 | INFO     | main | 性能指标: inference_latency_ms = 32.95850002905354
+2026-02-02 16:24:38 | INFO     | main | 性能指标: inference_latency_ms = 24.88639997318387
+2026-02-02 16:24:39 | INFO     | main | 性能指标: inference_latency_ms = 38.06369996163994
+2026-02-02 16:24:39 | INFO     | main | 性能指标: inference_latency_ms = 26.32400003494695
+2026-02-02 16:24:40 | INFO     | main | 性能指标: inference_latency_ms = 33.58839999418706
+2026-02-02 16:24:40 | INFO     | main | 性能指标: inference_latency_ms = 24.08180001657456
+2026-02-02 16:24:40 | INFO     | main | 性能指标: inference_latency_ms = 24.257299955934286
+2026-02-02 16:24:41 | INFO     | main | 性能指标: inference_latency_ms = 28.83950003888458
+2026-02-02 16:24:41 | INFO     | main | 性能指标: inference_latency_ms = 34.82649999205023
+2026-02-02 16:24:42 | INFO     | main | 性能指标: inference_latency_ms = 31.570899998769164
+2026-02-02 16:24:42 | INFO     | main | 性能指标: inference_latency_ms = 32.29719999944791
+2026-02-02 16:24:42 | INFO     | main | 性能指标: inference_latency_ms = 27.527199999894947
+2026-02-02 16:24:43 | INFO     | main | 性能指标: inference_latency_ms = 46.21130001032725
+2026-02-02 16:24:43 | INFO     | main | 性能指标: inference_latency_ms = 31.36849997099489
+2026-02-02 16:24:44 | INFO     | main | 性能指标: inference_latency_ms = 31.95969999069348
+2026-02-02 16:24:44 | INFO     | main | 性能指标: inference_latency_ms = 39.06999999890104
+2026-02-02 16:24:44 | INFO     | main | 性能指标: inference_latency_ms = 31.84859995963052
+2026-02-02 16:24:45 | INFO     | main | 性能指标: inference_latency_ms = 22.663299983832985
+2026-02-02 16:24:45 | INFO     | main | 性能指标: inference_latency_ms = 27.797500020824373
+2026-02-02 16:24:46 | INFO     | main | 性能指标: inference_latency_ms = 34.23879999900237
+2026-02-02 16:24:46 | INFO     | main | 性能指标: inference_latency_ms = 31.702900014352053
+2026-02-02 16:24:46 | INFO     | main | 性能指标: inference_latency_ms = 33.701500040479004
+2026-02-02 16:24:47 | INFO     | main | 性能指标: inference_latency_ms = 33.07240002322942
+2026-02-02 16:24:47 | INFO     | main | 性能指标: inference_latency_ms = 33.464199979789555
+2026-02-02 16:24:48 | INFO     | main | 性能指标: inference_latency_ms = 19.48159997118637
+2026-02-02 16:24:48 | INFO     | main | 性能指标: inference_latency_ms = 20.6579000223428
+2026-02-02 16:24:48 | INFO     | main | 性能指标: inference_latency_ms = 20.6447999808006
+2026-02-02 16:24:49 | INFO     | main | 性能指标: inference_latency_ms = 20.285000035073608
+2026-02-02 16:24:49 | INFO     | main | 性能指标: inference_latency_ms = 23.07879994623363
+2026-02-02 16:24:50 | INFO     | main | 性能指标: inference_latency_ms = 31.669999996665865
+2026-02-02 16:34:22 | INFO     | main | Edge_Inference_Service 初始化开始
+2026-02-02 16:34:22 | INFO     | main | ==================================================
+2026-02-02 16:34:22 | INFO     | main | Edge_Inference_Service 启动
+2026-02-02 16:34:22 | INFO     | main | ==================================================
+2026-02-02 16:34:22 | INFO     | main | 数据库初始化成功
+2026-02-02 16:34:27 | INFO     | main | 配置管理器初始化成功
+2026-02-02 16:34:27 | INFO     | main | 流管理器初始化成功
+2026-02-02 16:34:27 | INFO     | main | Batch预处理器: batch=1, target_size=(480, 480), fp16=True
+2026-02-02 16:34:27 | INFO     | main | 图像预处理器初始化完成: 输入尺寸 480x480, Batch大小 1, FP16模式 True
+2026-02-02 16:34:27 | INFO     | main | 预处理器初始化成功
+2026-02-02 16:34:27 | INFO     | main | TensorRT 引擎初始化: ./models/yolo11n.engine, batch=1, shape=(3, 480, 480), fp16=True
+2026-02-02 16:34:27 | INFO     | main | 连接事件: load - TensorRT -> ./models/yolo11n.engine
+2026-02-02 16:34:27 | INFO     | main | TensorRT 引擎加载成功: ./models/yolo11n.engine
+2026-02-02 16:34:27 | INFO     | main |   输入: 1, 输出: 1, batch=1
+2026-02-02 16:34:27 | INFO     | main | 引擎已加载: default
+2026-02-02 16:34:27 | INFO     | main | 推理引擎加载成功: ./models/yolo11n.engine
+2026-02-02 16:34:27 | INFO     | main | 后处理器初始化完成: NMS阈值=0.45, 置信度阈值=0.5
+2026-02-02 16:34:27 | INFO     | main | 后处理器初始化成功
+2026-02-02 16:34:27 | INFO     | main | 结果上报器初始化成功
+2026-02-02 16:34:31 | INFO     | main | 算法管理器初始化成功
+2026-02-02 16:34:31 | INFO     | main | 所有组件初始化完成
+2026-02-02 16:34:31 | INFO     | main | 视频流已添加: test_camera_01
+2026-02-02 16:34:31 | INFO     | main | 已添加摄像头: test_camera_01
+2026-02-02 16:34:35 | INFO     | main | 连接事件: connect - RTSP -> test_camera_01
+2026-02-02 16:34:35 | INFO     | main | 视频流已启动: test_camera_01 @ 5fps
+2026-02-02 16:34:35 | INFO     | main | 已启动 1 个视频流
+2026-02-02 16:34:35 | INFO     | main | Edge_Inference_Service 已启动
+2026-02-02 16:34:36 | INFO     | main | 性能指标: inference_latency_ms = 20.76500002294779
+2026-02-02 16:34:36 | INFO     | main | 性能指标: inference_latency_ms = 20.86390001932159
+2026-02-02 16:34:36 | INFO     | main | 性能指标: inference_latency_ms = 21.934199961833656
+2026-02-02 16:34:37 | INFO     | main | 性能指标: inference_latency_ms = 21.037899947259575
+2026-02-02 16:34:37 | INFO     | main | 性能指标: inference_latency_ms = 24.072199943475425
+2026-02-02 16:34:37 | INFO     | main | 性能指标: inference_latency_ms = 24.669900012668222
+2026-02-02 16:34:38 | INFO     | main | 性能指标: inference_latency_ms = 22.396799991838634
+2026-02-02 16:34:38 | INFO     | main | 性能指标: inference_latency_ms = 28.335100039839745
+2026-02-02 16:34:39 | INFO     | main | 性能指标: inference_latency_ms = 22.276600007899106
+2026-02-02 16:34:39 | INFO     | main | 性能指标: inference_latency_ms = 27.521800016984344
+2026-02-02 16:34:39 | INFO     | main | 性能指标: inference_latency_ms = 35.87540000444278
+2026-02-02 16:34:40 | INFO     | main | 性能指标: inference_latency_ms = 23.90420000301674
+2026-02-02 16:34:40 | INFO     | main | 性能指标: inference_latency_ms = 36.27240000059828
+2026-02-02 16:34:41 | INFO     | main | 性能指标: inference_latency_ms = 30.69159999722615
+2026-02-02 16:34:41 | INFO     | main | 性能指标: inference_latency_ms = 23.054699995554984
+2026-02-02 16:34:41 | INFO     | main | 性能指标: inference_latency_ms = 25.182199955452234
+2026-02-02 16:34:42 | INFO     | main | 性能指标: inference_latency_ms = 25.90119995875284
+2026-02-02 16:34:42 | INFO     | main | 性能指标: inference_latency_ms = 20.327099948190153
+2026-02-02 16:34:42 | INFO     | main | 性能指标: inference_latency_ms = 40.589299984276295
+2026-02-02 16:34:43 | INFO     | main | 性能指标: inference_latency_ms = 36.35279997251928
+2026-02-02 16:34:43 | INFO     | main | 性能指标: inference_latency_ms = 33.70070003438741
+2026-02-02 16:34:44 | INFO     | main | 性能指标: inference_latency_ms = 26.435399951878935
+2026-02-02 16:34:44 | INFO     | main | 性能指标: inference_latency_ms = 25.692099996376783
+2026-02-02 16:34:44 | INFO     | main | 性能指标: inference_latency_ms = 28.54179998394102
+2026-02-02 16:34:45 | INFO     | main | 性能指标: inference_latency_ms = 25.0998999690637
+2026-02-02 16:34:45 | INFO     | main | 性能指标: inference_latency_ms = 26.1859999736771
+2026-02-02 16:34:46 | INFO     | main | 性能指标: inference_latency_ms = 37.661599984858185
+2026-02-02 16:34:46 | INFO     | main | 性能指标: inference_latency_ms = 22.924199991393834
+2026-02-02 16:34:46 | INFO     | main | 性能指标: inference_latency_ms = 39.95710000162944
+2026-02-02 16:34:47 | INFO     | main | 性能指标: inference_latency_ms = 29.787199979182333
+2026-02-02 16:34:47 | INFO     | main | 性能指标: inference_latency_ms = 22.39870000630617
+2026-02-02 16:34:48 | INFO     | main | 性能指标: inference_latency_ms = 20.819300028961152
diff --git a/main.py b/main.py
index 920073d..3c7b270 100644
--- a/main.py
+++ b/main.py
@@ -190,10 +190,12 @@ class EdgeInferenceService:
             
             self._performance_stats["total_frames_processed"] += 1
             
-            self._logger.log_inference_latency(
-                processing_time_ms,
-                batch_size=1
-            )
+            self._frame_counter = getattr(self, '_frame_counter', 0) + 1
+            if self._frame_counter % 10 == 0:
+                self._logger.log_inference_latency(
+                    processing_time_ms,
+                    batch_size=1
+                )
             
         except Exception as e:
             self._logger.error(f"处理帧失败 {camera_id}: {e}")