Trained Faster-RCNN model with 200 classes of LSVRC detection competition - caffe

[Michele Svanera]

Hello to everyone!

For my research purposes, I'm looking for a faster-RCNN, already trained with 200 classes of LSVRC detection competition.

Could someone suggest me where to find an already-trained model, or even better share it with me?

Thanks!

[Michele Svanera]

Any suggestions?

[Lera Sh]

did you find this? 

[Zhaopeng Xuan]

Hi Michele,

May I ask you if for my dataset, which only have 1 class, I also only detect one object in the picture, how could I do that?

Thanks a lot.

Michele Svanera於 2016年2月26日星期五 UTC+1下午3時47分45秒寫道：

[Alessandro Ferrari]

Not exactly what you were asking for, however you may find useful:
https://github.com/weiliu89/caffe/tree/ssd
Faster prediction time, comparable mAP, no need for bouding boxes AND a pretrained model for MSCOCO 500.

Il giorno venerdì 26 febbraio 2016 15:47:45 UTC+1, Michele Svanera ha scritto:

[Zhaopeng Xuan]

Yo! Sorry... the question I posted yesterday is not clear .... maybe too tired yesterday

But what you proposed is really interesting! I'd like to try it for real-time prediction

Thanks 

Alessandro Ferrari於 2016年8月30日星期二 UTC+2上午10時13分18秒寫道：