X3D_Feature_Extraction

This repository contains code to extract **video features using the x3d_m model, given a folder of input videos.

⚠️ This code is intended for research purposes .

It can be used in the context of the following paper:

Weakly-supervised Video Anomaly Detection with Robust Temporal Feature Magnitude Learning

---

📦 Overview

This code processes each video in a folder and saves a corresponding NumPy file containing its extracted features.

• Input: A folder of videos or subfolders of videos
• Output: .npy feature file for each video
• Feature shape: (n/16, x, 2048) where n is the number of frames

---

🔗 Credits

This project is based on the following repositories:

• I3D Feature Extraction with ResNet
• I3D Feature Extraction with ResNet-50

✅ The backbone has been replaced with x3d_m in this version.

---

⚙️ Setup

Install dependencies:

pip install -r requirements.txt

Parameters

--datasetpath:       folder of input videos (contains videos or subdirectories of videos)
--outputpath:        folder of extracted features
--frequency:         how many frames between adjacent snippet
--batch_size:        batch size for snippets

Run

python main.py --datasetpath=dataset_path/ --outputpath=output

put the video in a folder and add the path to that folder in the datasetpath