Stockwell
A cashierless store ecosystem. At Stockwell, I designed, trained and deployed deep-learning models performing computer vision tasks to create receipts for customers from transaction videos recorded from the cameras in the store.
Anomaly Detection from Videos
I did my master's research at CIM, McGill University in the Intelligent Multimodal Video Analysis lab. During this time, I was able to propose a novel algorithm for generalized anomaly detection from videos using convolutional autoencoders by adaptive clustering in the latent space. A fully customizable, variable-length, open-source synthetic dataset generator was developed along with this work.
Image Inpainting
As part of the Deep-Learning course at Université de Montréal (IFT6266), I trained a generative deep-learning model (DC-GAN) that was able to fill in the missing parts of an image using the incomplete image and the image caption.
Can I Park Here?
It can be hard for people who do not speak French in the city of Montreal to find a parking spot since the parking signs are all in French. To try and solve this problem, I created an algorithm that will help them out.
