Alexander Quessy

icon indicating a website link https://aos55.github.io icon indicating an email [email protected]

icon company Bristol University icon location London, United Kingdom icon location University of Bristol PhD Student researching Deep Reinforcement Learning and Machine Learning for Aerial Robotics

Results 14 issues of Alexander Quessy

Consider Categories to use for the classifier

To continue to expand the computational complexity of the classifier will need to add more categories to be logged. This goes with expanding the dataset #18 to allow for the...

Add support for PyTorch profiler

This is the blog for the new [PyTorch profiller](https://pytorch.org/blog/introducing-pytorch-profiler-the-new-and-improved-performance-tool/), has vscode integration too which could be useful.

Support several convnets

2 comment

Support for the following ConvNets to compare the performance of each in various tasks: - [x] deeplabV3 - [x] deeplabV3+ - [x] Unet++ - [ ] HRNet

Segmentation Metrics

1 comment

Would be useful to support metrics to compare performance of various DCNNs in segmentation these could include: - [x] Number of network parameters - [ ] GFlops of compute needed...