Joonghee's Laboratory

https://towardsdatascience.com/illustrated-guide-to-lstms-and-gru-s-a-step-by-step-explanation-44e9eb85bf21

ZDNet: How do neural networks see depth?. https://www.zdnet.com/article/how-do-neural-networks-see-depth/

https://youtu.be/PahF2hZM6cs Watch "Recurrent Neural Networks (RNN) | RNN LSTM | Deep Learning Tutorial | Tensorflow Tutorial | Edureka" on YouTube https://youtu.be/y7qrilE-Zlc

https://www.google.com/url?sa=t&source=web&rct=j&url=https://www.slideshare.net/mobile/sunggonSong/8-75266382&ved=2ahUKEwjSz_qtgqXiAhXRMd4KHfNhCJwQFjADegQIBhAB&usg=AOvVaw0YQuTbjNUb8RUe9K4-APZ1&cshid=1558180708787

Alexnet VGGNet GooLeNet(Inception v1) ResNet

1st image. -> Zero padding : 1 2nd image. -> Zero padding : 2 http://deeplearning.net/software/theano/tutorial/conv_arithmetic.html

You should note P*. It's applied to both cls and reg. You can check it at the previous posting.

Answer to How does the region proposal network (RPN) in Faster R-CNN work? by Vahid Mirjalili https://www.quora.com/How-does-the-region-proposal-network-RPN-in-Faster-R-CNN-work/answer/Vahid-Mirjalili-1?ch=99&share=6f1f6a70&srid=u6gRt

https://www.slideshare.net/ssuser06e0c5/convolution-77257148 1. Common convolution input feature map : F x F x N(number of channels) output feature map : F x F x M(number of channels) => requirement: numbers of kernel parameters : K x K x N x M(number of kernels) total calculations : F x F x N x K x K x M 2. Pointwise convolution input feature map : F x F x N(number of chan..

https://github.com/pjreddie/darknet/issues/568