Overview
Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore advanced video compression techniques through pixel shift tracking in this 25-minute conference talk from Conf42 ML 2025. Learn about the evolution from traditional compression methods to AI-based approaches, with a focus on innovative pixel point tracking methodologies. Discover how single point trajectory and multi-point trajectory methods work to optimize video compression efficiency. Follow the complete compression and decompression workflow, analyze performance results, and understand future directions in video compression technology. Gain insights into how machine learning is revolutionizing video compression through sophisticated pixel tracking algorithms that improve compression ratios while maintaining video quality.
Syllabus
00:00 Introduction and Speaker Background
00:50 Overview of Video Compression
02:05 Traditional vs. AI-Based Compression Methods
04:18 Pixel Point Tracking Approach
06:28 Single Point Trajectory Method
09:08 Multi-Point Trajectory Method
11:18 Compression and Decompression Steps
17:08 Results and Performance
22:47 Future Approaches and Conclusion
Taught by
Conf42