YOLOv6
YOLOv6 copied to clipboard
mAP is low and no detection result.
Hi, using your own data for training, after 180 epochs, the accuracy is relatively low, and the inference has no result.
Hi, which dataset are you using?
I also met the problem that all the mAP is low,but yolov5lite is normal
Does it train normally with yolov5?
With the latest repo the mAP results on my small target dataset are still not on par with a v5 model of the same scale and the training is extremely unstable, but the difference is also not that big.
It could also be inappropriate hyperparameters. But we need to see more information from the training process to get an idea about where to tune.
@mtjhl Thank you for your reply, i use your own datasets, not a public dataset.
Hi, which dataset are you using?
i use the widerface and myself own datasets,reduce the flops to 60M,The accuracy of the validation set is very poor:
Inferencing model in val datasets.: 100%|##############################################################################| 64/64 [01:19<00:00, 1.24s/it]
Evaluating speed.
Evaluating mAP by pycocotools. Saving ./runs/train/yolov6_face_66M_0705/predictions.json... loading annotations into memory... Done (t=0.11s) creating index... index created! Loading and preparing results... DONE (t=2.71s) creating index... index created! Running per image evaluation... Evaluate annotation type bbox DONE (t=36.83s). Accumulating evaluation results... DONE (t=5.91s). Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.001 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.001 Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.001 Epoch: 60 | [email protected]: 0.00012148454109214603 | [email protected]:0.95: 1.989929560401417e-05
拉取一下最新的代码,在非coco 数据上,把这个 is_coco 设置为 False https://github.com/meituan/YOLOv6/blob/fb9485fdb3c5e290c51b8c7dc8b05e2bc78a8da3/data/coco.yaml#L9
拉取一下这个最新的代码,在非coco数据上,把is_coco设置为False
https://github.com/meituan/YOLOv6/blob/fb9485fdb3c5e290c51b8c7dc8b05e2bc78a8da3/data/coco.yaml#L9
试过了.
hi @sssssshf If dataset has been trained with yolov5/7 or older version of yolov6, the dataset cache might be incompatible for yolov6. Add --check-images and --check-images to create new cache.