train normally and generate output/trained_weights , but test_dota.py cannot detect any targets

[hopesala]

Hi 大佬, I can train my own data normally and generate output/trained_weights accordingly using the docker provided, but test_dota.py cannot detect any target. What could be the reason? Is my training process too short? or my datsat is too small? or the output/trained_weights is not used for detection? I cannot think a reason? can you give me some advice to debug this problem

My train command is RotationDetection-main/tools/r3det# python train.py

My detection command is /RotationDetection-main/tools/r3det# python test_dota.py --test_dir='/Downloads/RotationDetection-main/DOTA/test/' --gpus=0 -s --show_box

[hopesala]

I have vertified that the above problem was caused by too little training time, training for one day is ok to detect another question is below in file test_dota_base.py threshold = {'roundabout': 0.1, 'tennis-court': 0.3, 'swimming-pool': 0.1, 'storage-tank': 0.2, 'soccer-ball-field': 0.3, 'small-vehicle': 0.2, 'ship': 0.2, 'plane': 0.3, 'large-vehicle': 0.1, 'helicopter': 0.2, 'harbor': 0.0001, 'ground-track-field': 0.3, 'bridge': 0.0001, 'basketball-court': 0.3, 'baseball-diamond': 0.3, 'container-crane': 0.05, 'airport': 0.5, 'helipad': 0.1}

how to set the number after each type name reasonably, according to times occurred or width height ratio of the box? or anything else? thanks