← Computer Vision learning pathComputer Vision · Module 8 of 10
Video analysis
Process video streams for tracking and action recognition.
📖 Key ideas
- Video is a sequence of frames (e.g. 30 fps).
- Object tracking follows objects across frames (SORT, DeepSORT, ByteTrack).
- Optical flow measures motion between frames.
- Action recognition classifies activities.
- Real-time needs efficient models and hardware.
💡 Real-life example: Tracking is following one runner through a crowded marathon from frame to frame.
FrameSingle image in a video
fpsFrames per second
Object trackingFollowing objects across frames
Optical flowPixel motion between frames