how to visualize?

[zhouzheng1123]

How do you visualize the prediction image and draw the boxs on the image？

[naabdi]

Hello. I have this problem too. How can I visualize final boxes on the image?

[naabdi]

Have you reimplemented the results as shown in the paper ? @naabdi @zhouzheng1123

My results are slightly different with the paper and I think it is because of my hardware and my small input batch size. However I don't know how can I draw bounding boxes on the image.

[zhouzheng1123]

Have you reimplemented the results as shown in the paper ? @naabdi @zhouzheng1123

Have you reimplemented the results as shown in the paper ? @naabdi @zhouzheng1123

My results are slightly different with the paper and I think it is because of my hardware and my small input batch size. However I don't know how can I draw bounding boxes on the image.

me too

[zhouzheng1123]

I have visualized the bounding boxes on the image. @zhouzheng1123 @naabdi

can you please give us your email or open your code ？

[naabdi]

you can send your email address.

nariman_abdi[at]yahoo[dot]com

[zhouzheng1123]

you can send your email address.

[email protected]

[763593298]

@naabdi @zhouzheng1123 hi,can you please give me your code with visualize？Think you! my email is [email protected].

[DF416]

you can send your email address.

[email protected] Have you ever encountered the situation that the precision, recall50 and recall75 of each batch are 0 in base training?

[DF416]

Hello. I have this problem too. How can I visualize final boxes on the image?

Have you ever encountered the situation that the precision, recall50 and recall75 of each batch are 0 in base training?

[MusherM]

I've alreay finished the visualize code, I'll release it soon. Check it if you need it

[MusherM]

already commit to my fsodm fork:draw_bounding_box.py have fun with it

[Karblure]

Have you reimplemented the results as shown in the paper ? @naabdi @zhouzheng1123

My results are slightly different with the paper and I think it is because of my hardware and my small input batch size. However I don't know how can I draw bounding boxes on the image.

How long have you been training to achieve the same results as the paper? The Base Training is too long.

[wxhnibr]

@naabdi @zhouzheng1123 @MusherM @DF416 The results in the basic training are correct, but there are significant differences in the results in the new training. The accuracy of the aircraft class is particularly low. Can you tell me some details? Thank you very much！

[wxhnibr]

I trained 800 epochs, and the map for the base class is 0.78.which is similar to the results of the paper

[wxhnibr]

@Karblure I trained 800 epochs in base train, and the map for the base class is 0.78.which is similar to the results of the paper