vasgaowei / pytorch_MELM

The pytorch implementation of the Min-Entropy Latent Model for Weakly Supervised Object Detection
104 stars 19 forks source link

Reproducibility results #6

Closed zwyyy215 closed 5 years ago

zwyyy215 commented 5 years ago

Hello, thank you for your hard work. I would like to ask.What is the mAP of your experimental results?Is it the same as the result in the paper?

vasgaowei commented 5 years ago

My experiment's performance mAP is 47.98, trained in PASCAL_VOC 2007 trainval and tested on PASCAL_VOC test . The mAP is a little better than the papers' 47.3.

vasgaowei commented 5 years ago

And the backbone network is vgg16

zwyyy215 commented 5 years ago

好的,谢谢。想问问您,MELM的作者又出了一篇文章,C-MIL:Continuation Multiple Instance Learning for Weakly Supervised Object Detection这篇论文。因为这篇文章与MELM的作者是同一个人,但是这两篇文章的开源代码都是lua。我想问问,您还会整理CMIL的pytorch版本吗?