yolov5 int8量化推理结果为空 about tensorrtx HOT 6 OPEN

补充：
int8calib.table内容如下：
TRT-8601-EntropyCalibration2
data: 3b81b367
(Unnamed Layer* 0) [Convolution]_output: 3b817746
(Unnamed Layer* 1) [Scale]_output: 3d30e7d2
(Unnamed Layer* 2) [Activation]_output: 3af4a953
(Unnamed Layer* 3) [ElementWise]_output: 3d34ab0c
(Unnamed Layer* 4) [Convolution]_output: 3d3aa6d7
(Unnamed Layer* 5) [Scale]_output: 3e5e1936
(Unnamed Layer* 6) [Activation]_output: 3a420408
(Unnamed Layer* 7) [ElementWise]_output: 3e331811
(Unnamed Layer* 8) [Convolution]_output: 3d4bc6cb
(Unnamed Layer* 9) [Scale]_output: 3d89195b
(Unnamed Layer* 10) [Activation]_output: 3b174e9d
(Unnamed Layer* 11) [ElementWise]_output: 3cc4a61f
(Unnamed Layer* 12) [Convolution]_output: 3d27bbff
(Unnamed Layer* 13) [Scale]_output: 3e0cec3f
(Unnamed Layer* 14) [Activation]_output: 3a410204
(Unnamed Layer* 15) [ElementWise]_output: 3d3e8016
(Unnamed Layer* 16) [Convolution]_output: 3bea2522
(Unnamed Layer* 17) [Scale]_output: 3d71dd94
(Unnamed Layer* 18) [Activation]_output: 3b212245
(Unnamed Layer* 19) [ElementWise]_output: 3d2dad25
(Unnamed Layer* 20) [Convolution]_output: 3cf0b764
(Unnamed Layer* 21) [Scale]_output: 3d71021b
(Unnamed Layer* 22) [Activation]_output: 3ae20408
(Unnamed Layer* 23) [ElementWise]_output: 3d009a49
(Unnamed Layer* 24) [ElementWise]_output: 3d0e094e
(Unnamed Layer* 25) [Concatenation]_output: 3d5a066d
(Unnamed Layer* 26) [Convolution]_output: 3cb493c6
(Unnamed Layer* 27) [Scale]_output: 3d115893
(Unnamed Layer* 28) [Activation]_output: 3a5c3871
(Unnamed Layer* 29) [ElementWise]_output: 3cadd441
(Unnamed Layer* 30) [Convolution]_output: 3cd1e5cb
(Unnamed Layer* 31) [Scale]_output: 3cb90500
(Unnamed Layer* 32) [Activation]_output: 3bf48907
(Unnamed Layer* 33) [ElementWise]_output: 3c8d53fc
(Unnamed Layer* 34) [Convolution]_output: 3b959990
(Unnamed Layer* 35) [Scale]_output: 3c7bfde8
(Unnamed Layer* 36) [Activation]_output: 3ae6e9b5
(Unnamed Layer* 37) [ElementWise]_output: 3c6589e7
(Unnamed Layer* 38) [Convolution]_output: 3befb849
(Unnamed Layer* 39) [Scale]_output: 3c8b9b93
(Unnamed Layer* 40) [Activation]_output: 3bfed8cc
(Unnamed Layer* 41) [ElementWise]_output: 3c58f002
(Unnamed Layer* 42) [Convolution]_output: 3b61c65c
(Unnamed Layer* 43) [Scale]_output: 3c887834
(Unnamed Layer* 44) [Activation]_output: 3bf58b10
(Unnamed Layer* 45) [ElementWise]_output: 3cb0696b
(Unnamed Layer* 46) [Convolution]_output: 3c45a76e
(Unnamed Layer* 47) [Scale]_output: 3c4893a4
(Unnamed Layer* 48) [Activation]_output: 3bfe541f
(Unnamed Layer* 49) [ElementWise]_output: 3c25f7c7
(Unnamed Layer* 50) [ElementWise]_output: 3c8d4754
(Unnamed Layer* 51) [Convolution]_output: 3bf19c7b
(Unnamed Layer* 52) [Scale]_output: 3c8d01e3
(Unnamed Layer* 53) [Activation]_output: 3be5aa86
(Unnamed Layer* 54) [ElementWise]_output: 3c6ac1af
(Unnamed Layer* 55) [Convolution]_output: 3ce81f29
(Unnamed Layer* 56) [Scale]_output: 3d5b5c96
(Unnamed Layer* 57) [Activation]_output: 3c010a14
(Unnamed Layer* 58) [ElementWise]_output: 3cba5de5
(Unnamed Layer* 59) [ElementWise]_output: 3cc2b5a6
(Unnamed Layer* 60) [Concatenation]_output: 3d017324
(Unnamed Layer* 61) [Convolution]_output: 3c9577e6
(Unnamed Layer* 62) [Scale]_output: 3ca03324
(Unnamed Layer* 63) [Activation]_output: 3b816554
(Unnamed Layer* 64) [ElementWise]_output: 3bcb7117
(Unnamed Layer* 65) [Convolution]_output: 3bf94211
(Unnamed Layer* 66) [Scale]_output: 3d1c55c9
(Unnamed Layer* 67) [Activation]_output: 3bfad59b
(Unnamed Layer* 68) [ElementWise]_output: 3cd53e55
(Unnamed Layer* 69) [Convolution]_output: 3c1e1200
(Unnamed Layer* 70) [Scale]_output: 3c25d093
(Unnamed Layer* 71) [Activation]_output: 3bfd3475
(Unnamed Layer* 72) [ElementWise]_output: 3c2c32a1
(Unnamed Layer* 73) [Convolution]_output: 3c102c3f
(Unnamed Layer* 74) [Scale]_output: 3c916a80
(Unnamed Layer* 75) [Activation]_output: 3bff9f3c
(Unnamed Layer* 76) [ElementWise]_output: 3c815088
(Unnamed Layer* 77) [Convolution]_output: 3bb27a23
(Unnamed Layer* 78) [Scale]_output: 3d084cd5
(Unnamed Layer* 79) [Activation]_output: 3b81a31f
(Unnamed Layer* 80) [ElementWise]_output: 3d02e2dd
(Unnamed Layer* 81) [Convolution]_output: 3cf17b49
(Unnamed Layer* 82) [Scale]_output: 3c83603a
(Unnamed Layer* 83) [Activation]_output: 3bff1be9
(Unnamed Layer* 84) [ElementWise]_output: 3c73602a
(Unnamed Layer* 85) [ElementWise]_output: 3cbf77da
(Unnamed Layer* 86) [Convolution]_output: 3bc5373f
(Unnamed Layer* 87) [Scale]_output: 3d08f7d0
(Unnamed Layer* 88) [Activation]_output: 3b82213f
(Unnamed Layer* 89) [ElementWise]_output: 3c2cee2a
(Unnamed Layer* 90) [Convolution]_output: 3c5af977
(Unnamed Layer* 91) [Scale]_output: 3d058442
(Unnamed Layer* 92) [Activation]_output: 3bfd9abd
(Unnamed Layer* 93) [ElementWise]_output: 3d0406a3
(Unnamed Layer* 94) [ElementWise]_output: 3d08ed17
(Unnamed Layer* 95) [Convolution]_output: 3c50104e
(Unnamed Layer* 96) [Scale]_output: 3db08dda
(Unnamed Layer* 97) [Activation]_output: 3b81e3b0
(Unnamed Layer* 98) [ElementWise]_output: 3ce4f1bf
(Unnamed Layer* 99) [Convolution]_output: 3d02f014
(Unnamed Layer* 100) [Scale]_output: 3cf6f45c
(Unnamed Layer* 101) [Activation]_output: 3c0356a9
(Unnamed Layer* 102) [ElementWise]_output: 3c98a0f9
(Unnamed Layer* 103) [ElementWise]_output: 3d2331cf
(Unnamed Layer* 104) [Concatenation]_output: 3ce7e58b
(Unnamed Layer* 105) [Convolution]_output: 3c980c53
(Unnamed Layer* 106) [Scale]_output: 3db4f892
(Unnamed Layer* 107) [Activation]_output: 3be5c7fd
(Unnamed Layer* 108) [ElementWise]_output: 3c7fda5b
(Unnamed Layer* 109) [Convolution]_output: 3cd3c712
(Unnamed Layer* 110) [Scale]_output: 3ce3bb70
(Unnamed Layer* 111) [Activation]_output: 3ae74d63
(Unnamed Layer* 112) [ElementWise]_output: 3c9c74e2
(Unnamed Layer* 113) [Convolution]_output: 3c585200
(Unnamed Layer* 114) [Scale]_output: 3c5e64b9
(Unnamed Layer* 115) [Activation]_output: 3ae17813
(Unnamed Layer* 116) [ElementWise]_output: 3b8fa2ca
(Unnamed Layer* 117) [Convolution]_output: 3c35301c
(Unnamed Layer* 118) [Scale]_output: 3c29645d
(Unnamed Layer* 119) [Activation]_output: 3bffd8f5
(Unnamed Layer* 120) [ElementWise]_output: 3c43f736
(Unnamed Layer* 121) [Convolution]_output: 3b669cb8
(Unnamed Layer* 122) [Scale]_output: 3dd2548e
(Unnamed Layer* 123) [Activation]_output: 3b81a325
(Unnamed Layer* 124) [ElementWise]_output: 3c8e3e87
(Unnamed Layer* 125) [Convolution]_output: 3d52782a
(Unnamed Layer* 126) [Scale]_output: 3d377f3d
(Unnamed Layer* 127) [Activation]_output: 3c013254
(Unnamed Layer* 128) [ElementWise]_output: 3c9a53d9
(Unnamed Layer* 129) [ElementWise]_output: 3cc2cb29
(Unnamed Layer* 130) [Concatenation]_output: 3c73cbb5
(Unnamed Layer* 131) [Convolution]_output: 3ceb39f2
(Unnamed Layer* 132) [Scale]_output: 3cd4856a
(Unnamed Layer* 133) [Activation]_output: 3c0c4884
(Unnamed Layer* 134) [ElementWise]_output: 3caaf32d
(Unnamed Layer* 135) [Convolution]_output: 3d1cfed7
(Unnamed Layer* 136) [Scale]_output: 3d278595
(Unnamed Layer* 137) [Activation]_output: 3c00cfdc
(Unnamed Layer* 138) [ElementWise]_output: 3d2695b0
(Unnamed Layer* 139) [Pooling]_output: 3d2695b0
(Unnamed Layer* 140) [Pooling]_output: 3d2695b0
(Unnamed Layer* 141) [Pooling]_output: 3d2695b0
(Unnamed Layer* 142) [Concatenation]_output: 3d2695b0
(Unnamed Layer* 143) [Convolution]_output: 3d93b860
(Unnamed Layer* 144) [Scale]_output: 3c25d862
(Unnamed Layer* 145) [Activation]_output: 3ae1aabe
(Unnamed Layer* 146) [ElementWise]_output: 3c1240e8
(Unnamed Layer* 147) [Convolution]_output: 3cb7f33d
(Unnamed Layer* 148) [Scale]_output: 3d132964
(Unnamed Layer* 149) [Activation]_output: 3c02f1cb
(Unnamed Layer* 150) [ElementWise]_output: 3c5208a2
(Unnamed Layer* 151) [Resize]_output: 3c5208a2
(Unnamed Layer* 152) [Concatenation]_output: 3c327dc1
(Unnamed Layer* 153) [Convolution]_output: 3cda9571
(Unnamed Layer* 154) [Scale]_output: 3d35e030
(Unnamed Layer* 155) [Activation]_output: 3c022677
(Unnamed Layer* 156) [ElementWise]_output: 3d01f992
(Unnamed Layer* 157) [Convolution]_output: 3c8a2274
(Unnamed Layer* 158) [Scale]_output: 3cd5b648
(Unnamed Layer* 159) [Activation]_output: 3bfadbc4
(Unnamed Layer* 160) [ElementWise]_output: 3c14f50e
(Unnamed Layer* 161) [Convolution]_output: 3c3559ad
(Unnamed Layer* 162) [Scale]_output: 3db41fe1
(Unnamed Layer* 163) [Activation]_output: 3b821766
(Unnamed Layer* 164) [ElementWise]_output: 3d119652
(Unnamed Layer* 165) [Convolution]_output: 3d278d3b
(Unnamed Layer* 166) [Scale]_output: 3ce61cff
(Unnamed Layer* 167) [Activation]_output: 3c0b26a8
(Unnamed Layer* 168) [ElementWise]_output: 3c52b179
(Unnamed Layer* 169) [Concatenation]_output: 3c52b179
(Unnamed Layer* 170) [Convolution]_output: 3cb9b628
(Unnamed Layer* 171) [Scale]_output: 3d0f55b3
(Unnamed Layer* 172) [Activation]_output: 3bfd3c1c
(Unnamed Layer* 173) [ElementWise]_output: 3c407983
(Unnamed Layer* 174) [Convolution]_output: 3ca78e50
(Unnamed Layer* 175) [Scale]_output: 3d055ea8
(Unnamed Layer* 176) [Activation]_output: 3c0bcff4
(Unnamed Layer* 177) [ElementWise]_output: 3cd61402
(Unnamed Layer* 178) [Resize]_output: 3cd61402
(Unnamed Layer* 179) [Concatenation]_output: 3cd61402
(Unnamed Layer* 180) [Convolution]_output: 3d05768c
(Unnamed Layer* 181) [Scale]_output: 3d405449
(Unnamed Layer* 182) [Activation]_output: 3c0209b3
(Unnamed Layer* 183) [ElementWise]_output: 3ce8b506
(Unnamed Layer* 184) [Convolution]_output: 3c930084
(Unnamed Layer* 185) [Scale]_output: 3d19d063
(Unnamed Layer* 186) [Activation]_output: 3c183dda
(Unnamed Layer* 187) [ElementWise]_output: 3d1fab79
(Unnamed Layer* 188) [Convolution]_output: 3c781a8d
(Unnamed Layer* 189) [Scale]_output: 3d96d708
(Unnamed Layer* 190) [Activation]_output: 3bf99766
(Unnamed Layer* 191) [ElementWise]_output: 3cd0bfa6
(Unnamed Layer* 192) [Convolution]_output: 3d4b2af2
(Unnamed Layer* 193) [Scale]_output: 3d0b9c29
(Unnamed Layer* 194) [Activation]_output: 3c2570c1
(Unnamed Layer* 195) [ElementWise]_output: 3ce9d354
(Unnamed Layer* 196) [Concatenation]_output: 3d14b51a
(Unnamed Layer* 197) [Convolution]_output: 3ca46006
(Unnamed Layer* 198) [Scale]_output: 3e0633af
(Unnamed Layer* 199) [Activation]_output: 3c010a14
(Unnamed Layer* 200) [ElementWise]_output: 3e18cc16
(Unnamed Layer* 201) [Convolution]_output: 3e87aa4e
(Unnamed Layer* 202) [Convolution]_output: 3e52f481
(Unnamed Layer* 203) [Scale]_output: 3da1b59c
(Unnamed Layer* 204) [Activation]_output: 3c065ecb
(Unnamed Layer* 205) [ElementWise]_output: 3c197a5c
(Unnamed Layer* 206) [Concatenation]_output: 3cd61402
(Unnamed Layer* 207) [Convolution]_output: 3c934eaa
(Unnamed Layer* 208) [Scale]_output: 3ca567a0
(Unnamed Layer* 209) [Activation]_output: 3bf3c2f3
(Unnamed Layer* 210) [ElementWise]_output: 3c10d408
(Unnamed Layer* 211) [Convolution]_output: 3cb47b1b
(Unnamed Layer* 212) [Scale]_output: 3ccfbb86
(Unnamed Layer* 213) [Activation]_output: 3bf28575
(Unnamed Layer* 214) [ElementWise]_output: 3c954f22
(Unnamed Layer* 215) [Convolution]_output: 3c54c526
(Unnamed Layer* 216) [Scale]_output: 3dc454d6
(Unnamed Layer* 217) [Activation]_output: 3b819c13
(Unnamed Layer* 218) [ElementWise]_output: 3ce7eed1
(Unnamed Layer* 219) [Convolution]_output: 3cfe3e00
(Unnamed Layer* 220) [Scale]_output: 3cdde422
(Unnamed Layer* 221) [Activation]_output: 3c19f632
(Unnamed Layer* 222) [ElementWise]_output: 3cb14135
(Unnamed Layer* 223) [Concatenation]_output: 3d0229c6
(Unnamed Layer* 224) [Convolution]_output: 3c520f4e
(Unnamed Layer* 225) [Scale]_output: 3de00db0
(Unnamed Layer* 226) [Activation]_output: 3c01d398
(Unnamed Layer* 227) [ElementWise]_output: 3dc33798
(Unnamed Layer* 228) [Convolution]_output: 3e35d166
(Unnamed Layer* 229) [Convolution]_output: 3dd0403d
(Unnamed Layer* 230) [Scale]_output: 3cec2ac8
(Unnamed Layer* 231) [Activation]_output: 3c325459
(Unnamed Layer* 232) [ElementWise]_output: 3cb107d2
(Unnamed Layer* 233) [Concatenation]_output: 3cb107d2
(Unnamed Layer* 234) [Convolution]_output: 3d03b045
(Unnamed Layer* 235) [Scale]_output: 3d058b53
(Unnamed Layer* 236) [Activation]_output: 3bfa4ed8
(Unnamed Layer* 237) [ElementWise]_output: 3c121254
(Unnamed Layer* 238) [Convolution]_output: 3cf39c46
(Unnamed Layer* 239) [Scale]_output: 3c9eecc4
(Unnamed Layer* 240) [Activation]_output: 3ae21c45
(Unnamed Layer* 241) [ElementWise]_output: 3c92ea2c
(Unnamed Layer* 242) [Convolution]_output: 3c3be8d4
(Unnamed Layer* 243) [Scale]_output: 3dca0d6a
(Unnamed Layer* 244) [Activation]_output: 3b819886
(Unnamed Layer* 245) [ElementWise]_output: 3ce76ea1
(Unnamed Layer* 246) [Convolution]_output: 3cf0541b
(Unnamed Layer* 247) [Scale]_output: 3d4f5948
(Unnamed Layer* 248) [Activation]_output: 3c123d3f
(Unnamed Layer* 249) [ElementWise]_output: 3cd7055c
(Unnamed Layer* 250) [Concatenation]_output: 3cd7055c
(Unnamed Layer* 251) [Convolution]_output: 3cd443f4
(Unnamed Layer* 252) [Scale]_output: 3e170d92
(Unnamed Layer* 253) [Activation]_output: 3b81f3e8
(Unnamed Layer* 254) [ElementWise]_output: 3e143b2d
(Unnamed Layer* 255) [Convolution]_output: 3dd80192
prob: 4037eabe

from tensorrtx.

尝试增加校准图片了吗

from tensorrtx.

尝试增加校准图片了吗

感谢大佬，我使用了1900张图像作为校准图像，但是并没有效果

from tensorrtx.

换大一点模型试试吧，比如yolov5x

另外可以尝试下其他量化工具，不用trt自带的量化

from tensorrtx.

有什么推荐的量化工具吗？能接入trt的

from tensorrtx.