November 13th, 2022
Few-shot image classification based on Prototypical Embedding and Cosine Transformer. Introducing a new cross-attention mechanism based on Cosine Similarity that enhance few-shot transformer-based methods significantly.
March 18th, 2021
Introduction to few-shot learning, problem formulation, previous approaches, and baseline methods.
October 25th, 2021
A comprehensive summarization of Transformer-based architectures for Computer Vision tasks, including image classification, object detection, segmentation, and Few-shot Learning.
September 21st, 2021
A modified version of Mask R-CNN based on Matterport’s version. Featuring: polygon annotating mask generation based on model prediction, k-fold cross-validation training.