V.Dara comments

Results 8 comments of


                                            V.Dara

It is possible to replace its backbone with a yolov9 backbone?

well that is what I'm trying to do but I'm facing some problems

I would like to ask how it uses YOLOv8 as the backbone in his code, and specifically in which Python file the forward computation for the backbone is performed?

It's use thrid party [mmyolo](https://github.com/open-mmlab/mmyolo) Path: configs/finetune_coco/yolo_world_v2_xl_vlpan_bn_2e-4_80e_8gpus_mask-refine_finetune_coco.py That what I think

Can't reproduce yolov10s mAP. Closeing mosaic=True make mAP even worse

same

The value is the same before and after calling `init_weights` of YOLOWorldDetector 2024/12/22 02:04:44 - mmengine - INFO - Auto resumed from the latest checkpoint None. 2024/12/22 02:04:44 - mmengine...

训练结果中的bbox_mAP_s一直为 0.00 ?

trying to check the dataset for me it's the code part that I try to modify myself

A problem has arisen：ValueError: Key img_path is not in available keys.

pip install albumentations==1.3.1

train normal but val map all 0

Have you found the solution yet ?