about trained weights
Opened this issue · 4 comments
I used your code to train my own data. The loss drops normally. But when I used the trained weights to detect, no detection can be made.
I also tried VOC dataset on this training code, but still there is no detection. Seems there are some issues in the training process.
I got the same question. The loss seemed very tiny (at begining it's 5.1 and quickly down to 0.5) which was beyond my expectation and there was no detection when testing. I trained on the COCO dataset. I suspect that the loss function has some issues to address. Hope for author's reply. Danke!
I used your code to train my own data. The loss drops normally. But when I used the trained weights to detect, no detection can be made.
I also tried VOC dataset on this training code, but still there is no detection. Seems there are some issues in the training process.
Hi,I have the same question,I also trained it in vocdataset,and the train loss is in 0.4-0.7.But,the detection result is nothing!Do you solve this problem?Thanks!
I got the same question. The loss seemed very tiny (at begining it's 5.1 and quickly down to 0.5) which was beyond my expectation and there was no detection when testing. I trained on the COCO dataset. I suspect that the loss function has some issues to address. Hope for author's reply. Danke!
Hi,do you sloved this problem?I have the same problem recently! I used the model to test the image,but the output is none!There was no detection bbox in the picture.Thanks!
I got the same question. The loss seemed very tiny (at begining it's 5.1 and quickly down to 0.5) which was beyond my expectation and there was no detection when testing. I trained on the COCO dataset. I suspect that the loss function has some issues to address. Hope for author's reply. Danke!
Hi,do you sloved this problem?I have the same problem recently! I used the model to test the image,but the output is none!There was no detection bbox in the picture.Thanks!