jeong-tae

I think using cluster labels are what you to want.

Hello you guys. I heard that Retina performance on VOC may 55~60mAP which is very lower than SSD, YOLO variations. Maybe... because average number of boxes on images? I hope...

I also wanna know performance on VOC07, 12 Anyone tried?

@zhuhaozh Hi, Is there any progress on this? I am also working on this and I think the t**raditional MI can not calculate correct MI** on more than 2-dimensional variables....

@zhuhaozh Sure, the MINE can compute multi-dimensional variable case as well. For traditional way, you have to use multivariate normal distribution pdf instead of uni-variate normal distribution pdf. I am...

thanks. you mean that replacing chn to lsvt will work, right? btw, how can I evaluate lsvt? should I submit it to icdar board?

@mxin262 can you provide other link than [this](https://pan.baidu.com/s/1j-zlH8SfmdTtH2OnuT9B7Q)? I don't have account for that... so I can't download it.

@mxin262 thank you. your replay is really helpful to me. I tried with a new link, and I got an error `json.decoder.JSONDecodeError: Unterminated string starting at: line 1 column 419200878...

@mxin262 I am sorry, it was not the problem of the LSVT json file. It was `chn_syn.json` problem. I don't know why it occurs, but I used this [link](https://drive.google.com/u/0/uc?id=199sLThD_1e0vtDmpWrAEtUJyleS8DDTv&export=download). md5sum...

@mxin262 nope, I can't ``` python 3.8 >> json.load(open('chn_syn.json')) ``` returns `json.decoder.JSONDecodeError: Unterminated string starting at: line 1 column 419200878 (char 419200877)` this error.