YOLO-x test results incoherent

[guillermodecelisrodriguez]

Thanks for the amazing job on your publication.

i am trying to replicate your results, but if I use the only yolo_x_model.pt that you uploaded to drive, the results from testing yolo dont make sense i dont know what i am missing.

Thanks in advance!

I am running >> ./det/yolox/tools/test_yolox.sh ./configs/yolox/bop_pbr/yolox_x_640_augCozyAAEhsv_ranger_30_epochs_ycbv_real_pbr_ycbv_bop_test.py 0 ./pretrained_models/yolox/yolox_x.pth

and i obtain the following results:

Per-category bbox AP:

category	AP	category	AP	category	AP
002_master_chef_can	0.000	003_cracker_box	0.000	004_sugar_box	0.000
005_tomato_soup_can	0.000	006_mustard_bottle	0.000	007_tuna_fish_can	0.000
008_pudding_box	0.001	009_gelatin_box	0.000	010_potted_meat_can	0.241
011_banana	0.054	019_pitcher_base	0.021	021_bleach_cleanser	0.039
024_bowl	0.000	025_mug	21.278	035_power_drill	4.161
036_wood_block	0.000	037_scissors	0.000	040_large_marker	0.000
051_large_clamp	0.008	052_extra_large_clamp	0.000	061_foam_brick	0.032

[Rainbowend]

You're testing with the pretrained model, use the trained model (in BaiDuYunPan) instead.