Implement Yolo

[HuguesDug]

Hello,

The library is realy moving forward on a lot of topics and gives more and more the ability to process CNN in an efficient way. Thanks for this huge work.

Could you incorporate an example of Yolo Net architecture inside the examples. Seeing the layer types that are now existing, I think a Yolo V1 structure is possible, even V2. For V3, they have a "route" layer type, that is not possible yet I guess. V4 is just released and also integrate a special concat layer. May be possible with your newly added "concatinputoutpout".

Thanks

[joaopauloschuler]

Thank you for the feed back!

This request made me realize that nowhere in the documentation we have explanation about how to build multi-path architectures. I'll add documentation explaining how to create multi-path architectures with CAI.

Found some implementations that could be ported from Keras:

YOLO V1:
https://github.com/FMsunyh/keras-yolo

YOLO V2:
https://github.com/qjadud1994/YOLOv2-keras/blob/master/Model.py

YOLO V3:
https://github.com/qqwweee/keras-yolo3/blob/master/yolo3/model.py

Have just realized that the documentation doesn't say how to port code from Keras to Pascal. I'll add this in the near future.

[DrDub]

From the Villavu/Simba#507 discussion, YOLO is a must have. If you can point me to some code you have ported from Keras to Neural API, I can give it a try, both the documentation and the YOLO porting. I'm very familiar with Keras.