Have you pretrained weights on COCO dataset?I only got 66.5 mAP,trained on voc0712,tested in voc07....

[ghost]

No description provided.

[ke123tw]

I have the same problem. I trained on voc0712,tested in voc07, but mAP only have 55.1%.

[321zhangli123]

Have you trained on COCO ? I followed the author's experiment settings, but AP50 on test-dev only have 31.6%, which is 40.4% in the paper, I am very confused about this.

[peterpaniff]

Have you trained on COCO ? I followed the author's experiment settings, but AP50 on test-dev only have 31.6%, which is 40.4% in the paper, I am very confused about this.

Do you have trained the model on COCO? let's discuss something

[peterpaniff]

No description provided.

do you have trained the model on COCO, let's chat

[peterpaniff]

I have the same problem. I trained on voc0712,tested in voc07, but mAP only have 55.1%.

do you have trained the model on COCO? Let's chat

[321zhangli123]

Have you trained on COCO ? I followed the author's experiment settings, but AP50 on test-dev only have 31.6%, which is 40.4% in the paper, I am very confused about this.

Do you have trained the model on COCO? let's discuss something

I didn't trained on COCO again, but I found the low mAP maybe because my training parameters were different from the author，such as batch_size and base_lr. you can discuss with the author.

[peterpaniff]

Have you trained on COCO ? I followed the author's experiment settings, but AP50 on test-dev only have 31.6%, which is 40.4% in the paper, I am very confused about this.

Do you have trained the model on COCO? let's discuss something

I didn't trained on COCO again, but I found the low mAP maybe because my training parameters were different from the author，such as batch_size and base_lr. you can discuss with the author.

Do you save the pretrained model on COCO? I can not train the model on COCO because of my limited computation source. Could you share it with me if you have the pretrained model? thank you very much and happy new year

[321zhangli123]

Have you trained on COCO ? I followed the author's experiment settings, but AP50 on test-dev only have 31.6%, which is 40.4% in the paper, I am very confused about this.

Do you have trained the model on COCO? let's discuss something

I didn't trained on COCO again, but I found the low mAP maybe because my training parameters were different from the author，such as batch_size and base_lr. you can discuss with the author.

Do you save the pretrained model on COCO? I can not train the model on COCO because of my limited computation source. Could you share it with me if you have the pretrained model? thank you very much and happy new year

I did save the model，but the result is what I mentioned above. If you need, I can send it to you by email.

[peterpaniff]

Have you trained on COCO ? I followed the author's experiment settings, but AP50 on test-dev only have 31.6%, which is 40.4% in the paper, I am very confused about this.

Do you have trained the model on COCO? let's discuss something

I didn't trained on COCO again, but I found the low mAP maybe because my training parameters were different from the author，such as batch_size and base_lr. you can discuss with the author.

Do you save the pretrained model on COCO? I can not train the model on COCO because of my limited computation source. Could you share it with me if you have the pretrained model? thank you very much and happy new year

I did save the model，but the result is what I mentioned above. If you need, I can send it to you by email.

I do need, thank you very much. my e-mail address is : 15102716105@163.com. Thank you again.

[kunalgoyal9]

Can you please send me at kunalgoyal.goyal9@gmail.com

Thanks

[chumingqian]

Hi, @321zhangli123 ，@peterpaniff ,
Can you share a baidu Yun link or Google drive for the coco .weights,
Thanks in advance