Related Experiment Video
Updated: Oct 8, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Human Segmentation and Tracking Survey on Masks for MADS Dataset
1Department of Information Technology, Tan Trao University, Tuyen Quang 22000, Vietnam.
Sensors (Basel, Switzerland)
|December 28, 2021
Summary
This study surveys human segmentation and tracking methods using Convolutional Neural Networks (CNNs). It introduces the MASK MADS dataset for evaluating these techniques on complex activities.
Area of Science:
- Computer Vision
- Artificial Intelligence
- Machine Learning
Background:
- Human segmentation and tracking heavily rely on accurate person detection in videos.
- Convolutional Neural Networks (CNNs) have shown significant advancements in these areas.
- Applications include video monitoring and human pose estimation in 2D and 3D.
Purpose of the Study:
- To conduct a comprehensive survey of human segmentation and tracking methods in videos.
- To analyze the impact of person detection on segmentation and tracking performance.
- To introduce a new mask dataset for evaluating these tasks.
Main Methods:
- Survey of existing literature, methods, datasets, and results for human segmentation and tracking.
- Detailed examination of Convolutional Neural Networks (CNNs) approaches.
- Creation and utilization of the MASK MADS dataset for evaluation.
Main Results:
- The survey provides an in-depth review, including source code paths.
- The MASK MADS dataset contains 28,000 mask images for segmentation and tracking evaluation.
- Recent CNNs methods were evaluated on the MADS dataset for segmentation and tracking.
Conclusions:
- Accurate person detection is crucial for effective human segmentation and tracking.
- The MASK MADS dataset offers a valuable resource for benchmarking segmentation and tracking algorithms.
- The study highlights the effectiveness of CNNs in complex human activity analysis.
