jester-v1

Developing a pipeline for Hand Gesture Recognition

Prepare Jester-v1 dataset

Segment-based gesture classification using ResNeXt-101 features and lightweight MLPs, trained on the Jester dataset. This project explores frame sampling strategies, MLP architectures, and segment counts to balance accuracy with computational efficiency. Includes training code, evaluation notebooks, logs, metrics, and full experimental report.

Comprehensive Analysis of the Jester Dataset Using State-of-the-Art Video Classification Models.