My most recent results are available! I implemented the SincNet in Pytorch! My results were logged using , an amazing visualisation tool for sweeps. The dataset is, for now, a third of the Egyptian FruitBat dataset. Soon, I will add other animal recordings in order to asset the generalisation performances of my network!

I will also compute the IoU for the best classes, and shall compare this implementation with my trainings on YOLOR (You Only Learn One Reprensentation) which I previously implemented.

I look forward to sharing more. My next task will be to implementd DENet, which is only available on tensorflow, and integrate transformers to my architectures.