Master Internship project
-
Updated
Feb 19, 2021 - Python
Master Internship project
AVT Dissertation Research, Architecture used for AVT experimentation, mainly with backbone model and attention heads.
Video Query System
In progress: Uses CNNs to classify different forms for various exercises through a PyTorch implementation of "2D/3D Pose Estimation and Action Recognition using Multitask Deep Learning."
Exploration of different solutions to action recognition in video, using neural networks implemented in PyTorch.
AI privacy tool for pure edge computing utilizing translation, transcription across Hebrew, English and Farsi, Summarization, NER, Action Item Recognition, Timeline Extraction, Sentiment Analysis and Recommendations of the uploaded file.
Conclusion of undergraduate research in NIP-UPD
Python script leveraging pre-trained ResNet18 for extracting video features from the YouTube Dataset, enabling LSTM-based action recognition models.
Using two stream architecture to implement a classic action recognition method on UCF101 dataset
VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training
Tool employed to visualize synchronized FrVD metadata and videos simultaneously.
You Only Watch Once: A Unified CNN Architecture for Real-Time Spatiotemporal Action Localization
Semantic communication framework achieving 82.5% accuracy with 14,400:1 compression for explainable action recognition
A OpenMMLAB toolbox for human pose estimation, skeleton-based action recognition, and action synthesis.
In this repository, we have the feature extraction code, This code uses ResNet50 to extract spatial features from video frames.
Фреймворк компьютерного зрения для задач, детекции, сегментации, классификации, идентификации людей, анализа настроений и классификации действий
Add a description, image, and links to the action-recognition topic page so that developers can more easily learn about it.
To associate your repository with the action-recognition topic, visit your repo's landing page and select "manage topics."