Beyond painting buttons: computer vision techniques
summary
A history-and-comparison talk about computer vision approaches before and alongside neural networks, covering classical techniques developers used to make machines 'see' and concrete problem-solving methods for mobile development. It explains when neural networks are outperformed by non‑learning methods and discusses practical criteria for choosing between classical CV algorithms and modern deep learning.
AI-written summary · about
chapters (14)
- 0:00Welcome
- 1:01Speaker intro
- 1:24What is computer vision
- 4:13Neural network approaches
- 9:54Pretrained models
- 10:14Classical approaches
- 13:54How QR codes work
- 16:31Neural networks vs classical
- 18:59Approach symbiosis
- 22:01Golf swing sync example
- 22:21Tennis table detection example
- 27:42Roulette ball prediction example
- 33:42Summary
- 34:03Q&A
related talks
14mA Primer to Modern Day Computer Vision
40mArtificial Intelligence in Computer Vision
37mFrom Academia to Industry: Computer Vision with Deep Learning
44mComputer Vision in Practice
37mComputer Vision for Coders
03mRethinking software engineering in the AI era
04mUsing Computer Vision to Understand Big Visual Data
21mRecognizing Human features using Deep Networks