Related Experiment Video
Updated: Jun 3, 2025
08:25
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
Published on: May 7, 2019
8.9K
Video-Based Plastic Bag Grabbing Action Recognition: A New Video Dataset and a Comparative Study of Baseline Models.
Pei Jing Low1, Bo Yan Ng1, Nur Insyirah Mahzan1
1NUS-ISS, National University of Singapore, Singapore 119615, Singapore.
Sensors (Basel, Switzerland)
|January 11, 2025
Summary
This study introduces a new dataset and evaluates three methods for recognizing plastic bag grabbing actions in CCTV footage. Convolutional Neural Networks (CNNs) show promise for this specialized computer vision task.
Area of Science:
- Computer Vision
- Machine Learning
- Action Recognition
Background:
- Action video classification is a complex field, with plastic bag grabbing being a niche challenge.
- Existing methods may not adequately capture the nuances of this specific action in real-world CCTV data.
Purpose of the Study:
- To develop a specialized benchmark dataset for plastic bag grabbing action recognition.
- To propose and evaluate baseline methods for this task, including handcrafted features, 2D CNNs, and 3D CNNs.
Main Methods:
- A novel benchmark dataset was created for plastic bag grabbing.
- Three approaches were evaluated: handcrafted features with sequential models, multiple-frame Convolutional Neural Networks (CNNs), and 3D CNNs.
Main Results:
- Comparative analysis of the three baseline methods was performed.
- Strengths and limitations of each approach in recognizing plastic bag grabbing actions were identified.
Conclusions:
- The study provides a foundation for future research in specialized action recognition.
- The developed dataset and evaluated methods contribute to advancing computer vision capabilities for CCTV analysis.
