features of the model output are not consistent with those in the TSV file

[zhangchenghua123]

 hello,When I use the weight of the file resnet101_faster_rcnn_final.caffemodel to generate the bottom up feature of 10 to 100 objects,The boxes I get don't match the ones in the TSV file, or even in the same amount。
 for example,the picture:COCO_val2014_000000349021.jpg
 in tsv file,the boxes are:

[[ 72.63866 323.50052 638.9267 477.20337 ]
[ 0. 0. 531.0018 160.37456 ]
[196.17493 202.92207 582.0358 392.5355 ]
[298.6358 5.273723 638.9267 147.76785 ]
[489.84888 322.74255 638.9267 466.3207 ]
[184.39835 30.255976 362.12308 214.42981 ]
[454.17618 277.0955 473.7012 312.25125 ]
[ 0. 0. 247.79265 249.6724 ]
[ 0. 381.52252 302.8443 477.20337 ]
[246.22423 258.31384 283.56186 318.43063 ]
[334.2851 338.47977 392.62964 392.6544 ]
[ 47.213596 182.49136 227.34457 248.61078 ]
[164.66795 0. 588.44946 261.22583 ]
[246.60812 93.04805 618.67426 248.01872 ]
[400.15796 336.73657 450.4243 380.4387 ]
[306.83762 328.26505 519.74115 393.41925 ]
[178.46791 20.020086 638.9267 154.33356 ]
[197.05843 52.80718 334.35858 147.37617 ]
[533.7555 229.08055 561.8308 268.14777 ]
[455.75137 327.14612 504.37048 371.77405 ]
[ 59.182163 282.64435 505.19406 421.88565 ]
[ 26.859505 388.26898 237.05745 461.26172 ]
[312.04608 0. 567.1143 198.15587 ]
[205.47354 25.19839 337.72504 236.4498 ]
[238.22566 407.087 471.11194 473.29214 ]]
(25, 4)

The output of the model is:
[[ 72.63848 323.50046 638.9267 477.20337 ]
[ 0. 0. 531.0004 160.37056 ]
[196.19057 202.92047 582.09283 392.55243 ]
[298.63583 5.273705 638.9267 147.76787 ]
[489.8488 322.7419 638.9267 466.3208 ]
[184.39828 30.255981 362.12308 214.42964 ]
[ 0. 0. 247.79306 249.67366 ]
[ 0. 381.52252 302.84518 477.20337 ]
[246.22432 258.31384 283.56177 318.43027 ]
[334.28506 338.47772 392.63065 392.65598 ]
[ 47.213715 182.49133 227.34451 248.61078 ]
[164.66783 0. 588.44934 261.22583 ]
[246.60817 93.04582 618.67664 248.02307 ]
[400.1564 336.7355 450.4238 380.44537 ]
[306.84656 328.25873 519.7152 393.42056 ]
[178.46797 20.020117 638.9267 154.33359 ]
[197.0584 52.80718 334.35864 147.3761 ]
[533.7553 229.0807 561.8305 268.14844 ]
[455.75162 327.14432 504.37112 371.7736 ]
[ 59.182575 282.64435 505.1954 421.88574 ]
[ 26.859007 388.26907 237.05798 461.2615 ]
[312.04608 0. 567.1144 198.15587 ]
[205.4735 25.198402 337.72507 236.45001 ]
[238.22583 407.08698 471.11166 473.29208 ]
[453.46585 277.5032 469.12708 307.4863 ]
[521.9229 299.64764 638.9267 477.20337 ]]
(26, 4)
Did I do something wrong?