dataset

[377fsq]

Can you provide image file structure of LLVIP？

[DocF]

You need to convert all annotations to YOLOv5 format.

Refer: https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data

[377fsq]

Thanks for your answer, I have finished the training. But there are some questions. I used the yolov5l_fusion_transformerx3_llvip.yaml to train with default parameters, but the training results did not reach the values shown in your paper. My result mAP50: 0.965 mAP75: 0.709 mAP: 0.624 GPU: 3090Ti batch_size: 16. I want to know whether add other training strategies?

[DocF]

First, my input image size is 1024×1024. Second, I tested thebest.pt.

[377fsq]

Thank you, I will try again.

[nineRishav]

Hi @377fsq ,

I need you help in running the project. I wanted to train, test the model. But, I am facing issue in mentioning the path of FLIR_alligned.

What should I enter in the 4 fields of train_rgb, val_rgb, train_ir, val_ir.