Open 121649982 opened 2 weeks ago
I have the same problem. I run "python main.py --mode detect --config config/ucf_config.yaml", and data_root=E:/user/ucf24, the pretrained model ema_epoch_7.pth is downloaded from C23. The saved images in detection results have no bounding boxes and labels.
I have the same problem. I run "python main.py --mode detect --config config/ucf_config.yaml", and data_root=E:/user/ucf24, the pretrained model ema_epoch_7.pth is downloaded from C23. The saved images in detection results have no bounding boxes and labels.
I noticed that there are a few bounding boxes on several images, but most of the images do not have bounding boxes.
@Jiwen97 Thank you for your detailed report. I just tested on another system, everything is still fine. There are a few things I need to confirm that you are doing correctly:
In each checkpoint folder, there is a corresponding config file included. When you want to use a checkpoint, make sure you have downloaded the corresponding config file for that checkpoint and pass the path of that config file into the command line (not the config file available in the repository!). Additionally, you will need to edit the "pretrain_path" parameter in the config file to match the path of the checkpoint file you are using. If there are still issues, feel free to ping me, I'm always here to support everyone :3
Please provide me with more context (such as the config file, pretraining details...) and some samples from your data.