能否将源码开源到mmdetection中？

[916968505]

感觉代码结构和mmdetection很像，是不是合并以后传播度会更广？感谢！

[yuantn]

感谢您的建议，只是MI-AOD的这个代码主要是主动学习（Active Learning）领域的，检测只是其应用的一个具体领域。换句话说，我的MI-AOD的代码对于提高全监督情况下目标检测的性能较为有限，且主要目的不在于此，反而mmdetection代码中都是一些像RetinaNet, Freeanchor, PAA等这种专门致力于提高检测性能的方法，所以感觉有些不太适宜。如果存在一些能够集成各种主动学习方法的代码库的话，我很愿意加入进去。谢谢！

…

--- Thank you for your suggestion, but this code of MI-AOD is mainly for Active Learning, and object detection is only an application. In other words, my MI-AOD is limited in improving the performance of object detection under full supervision, and it is not the main purpose. Instead, the mmdetection is dedicated to RetinaNet, Freeanchor, PAA, etc. These are all methods to improve the detection performance, so it is inappropriate to add MI-AOD into mmdetection. If there are some toolboxes that can integrate various active learning methods, I would be willing to join them. Thank you!

[yuantn]

感觉代码结构和mmdetection很像，是不是合并以后传播度会更广？感谢！

对了，上面是我的回复，之前忘记引用了。

---

By the way, the above is my reply, I forgot to quote it before.