GTC 2020: NVIDIA Tool to Visualize and Interact with Feature Maps

GTC 2020 D2S17
Presenters: Tech Demo Team,NVIDIA
This demo shares how you can leverage the powerful new tool named ‘NVIDIA Feature Map Explorer’ to visualize 4-dimensional image-based feature map data in a fluid and interactive fashion. The tool provides developers with a rich set of views into feature map data that range from high-level summary to low-level channel slices, as well as detailed statistics information. The visualization and statistics enable deep learning developers to easily peer into the deep learning processing “black box” to find intimate information about what the model is learning, where the model is failing to use resources efficiently, and what is changing as a model is learning during training.

Watch this session
Join in the conversation below.

Would anyone be willing to share a tensorflow/keras code snippet where
feature maps are dumped from a model ?

I still find the vocabulary confusing, and a working example would be very
much appreciated.

I guess nobody is …