PREDATOR - Tracker that learns and improves
Channel:Zdenek Kalal · Length:4:12 · Published:2011-01-01 · Views:892889
Summary
A demo of PREDATOR (TLD), a real-time object tracker that learns the appearance of its target and improves while it tracks. The technology was later developed commercially.
Why HN recommends it
HN found it a nifty demo and cites it as a primitive but effective learning tracker that predates modern deep-learning trackers.
Very nifty. — HN, 2011
Machine learning & deep learning · Computer vision, speech & NLP
Subtitles (excerpt)
hi my name is den kalal and I'm a researcher focusing on visual object tracking this red dot is the result of my PhD thesis where I investigated how to track an unknown object in a video sequence and how to improve the tracking performance during tracking um we have developed an algorithm which we call Predator as any other tracker Predator estimates the object location and scale changes but in addition it learns how the object looks and what is not an object and using this information it scans every frame searching for the object if the object disappears and later comes back we are able to detect it this combination of tracking learning and detection is unique and therefore opens completely new possibilities for a number of industrial applications first of them is in human computer interfaces consider you want to make an interface where the user catches and drags something such as a mouse uh so using our system it's possible to select uh the object which you want Track by a bounding box and that's it now the system is learning how the gesture looks like in different scales how it uh looks in different rotations and so on and after a while we have pretty reliable system which is acting as a virtual Mouse the nice property
On Hacker News
My friend forwarded me the video demo link for this: [VIDEO] Very nifty. I'm looking forward to reading the machine learning code involved in this. — tom_b, 2011-04
OK, 11 years ago then. This time with some very primitive ML: OpenTLD [VIDEO] Was good enough for commercialization [LINK] . I think their tech is powering this [LINK] — rasz, 2023-08
Incorrect. Recognition applications have been able to track objects across disparate frames of video even when obscures from camera view, for at least a decade now [1]. 1. [VIDEO] — jackcviers3, 2021-07
Related videos
- The spelled-out intro to neural networks and backpropagation: building micrograd · Andrej Karpathy · 2:25:52
- But what is a neural network? | Deep learning chapter 1 · 3Blue1Brown · 18:40
- AlphaGo - The Movie | Full award-winning documentary · Google DeepMind · 1:30:28
- Transformers, the tech behind LLMs | Deep Learning Chapter 5 · 3Blue1Brown · 27:14
- USENIX Security '18-Q: Why Do Keynote Speakers Keep Suggesting That Improving Security Is Possible? · USENIX · 51:22
- Lecture 1 | Machine Learning (Stanford) · Stanford · 1:08:40
- MarI/O - Machine Learning for Video Games · SethBling · 5:57
- Deep Dive into LLMs like ChatGPT · Andrej Karpathy · 3:31:23
- Computer program that learns to play classic NES games · suckerpinch · 16:17
- RL Course by David Silver - Lecture 1: Introduction to Reinforcement Learning · Google DeepMind · 1:28:12
- Let's build GPT: from scratch, in code, spelled out. · Andrej Karpathy · 1:56:20
- Bay Area Vision Meeting: Unsupervised Feature Learning and Deep Learning · Google TechTalks · 48:10
- A Hackers' Guide to Language Models · Jeremy Howard · 1:31:13
- "Voice Driven Development: Who needs a keyboard anyway?" by Emily Shea · Strange Loop Conference · 41:17
- Andrej Karpathy - AI for Full-Self Driving at Tesla · Matroid · 30:08
- [1hr Talk] Intro to Large Language Models · Andrej Karpathy · 59:48