Thats you so much for the Series Very clean explanation of the MobileNet v2 paper
@Induraj11
2 жыл бұрын
Became a Fan of your videos! Keep going! you are an born teacher!
@iimadalishah
2 жыл бұрын
Great explanation, Where can we find the source code for training and testing in Python or MATLAB?
@zardouayassir7359
2 жыл бұрын
API for Python/TensorFlow: www.tensorflow.org/api_docs/python/tf/keras/applications/mobilenet_v2 Good luck
@yeahno2466
10 ай бұрын
Can you please make a video like this but with MobileNetv3? 🥺🙏
@zardouayassir7359
10 ай бұрын
I appreciate that you wanted a video from my channel. My time is quite limited at the moment, but I'll definitely consider your request. Thanks for your understanding
@Idzi005
Жыл бұрын
This is a really great video. Thanks!
@nunooliveira9322
3 жыл бұрын
Great video. On 2:24 128 channels are related to 128 kernels? So the feature map has 128 images? Thank you
@zardouayassir7359
3 жыл бұрын
Hi Nuno, This is a quick answer: In most cases YES. Let's say that H, W, D1 are the dimensions of a feature map, which is input to a convolutional layer (ConvL) that produced a new feature map having the dimensions H, W, D2. The fact that the output feature map has D2 channels does not necessarily mean that the number of kernels in ConvL are equal to D2. If you use standard convolutions or pointwise convolutions, where the depth (i.e., number of channels) of all kernels is equal to the depth (i.e., number of channels) of the input feature map, then the number of convolutional kernels is equal to the depth of the output feature map D2. But if you use depthwise convolutions, where the depth of all kernels is equal to 1, then the number of convolutional kernels is not equal to the depth of the output feature map D2. Thank you for your kind comment!
@nunooliveira9322
3 жыл бұрын
@@zardouayassir7359 Thank you very much. Your videos are amazing.
Пікірлер: 14