Gorilla-Lab-SCUT/frustum-convnet

ERROR: Couldn't read: 004391.txt of ground truth.

Closed this issue · 4 comments

Hello,
zhixin, I have ran the codes follow your step. But I had a problem when I used the testing codes, which is as follows.
<python train/test_net_det.py --cfg cfgs/det_sample.yaml OUTPUT_DIR output/car_train TEST.WEIGHTS output/car_train/model_0050.pth>
An error occured :
ERROR: Couldn't read: 004391.txt of ground truth. Please write me an email!
An error occured while processing your results.

Could you please help me solve it?

please double check your dataset and your label path. Can you find label files by the path in your code?

@lhyfst Thank you for your reply!
And I have found that some of .txt files inoutput/car_train/val_nms/result/data are empty. Is there something wrong with my training,or do you know the reasons?

yes, some frames just do not have any cars. If you still suspect there are some problems, you had better visualize your prediction.

@lhyfst I've got it. Many thanks!