Week 2: Data Pipeline
Day 9: Frames and Timing
Goal
Understand why video becomes frame sequences.
Learn
- A video is a sequence of frames. Many pipelines process each frame, then study how the body, hands, and face change over time.
- Timing matters in ASL. Holds, repeated movement, speed, transitions, and pauses can all affect meaning or naturalness.
- Frame rate and trimming affect the data. A clip that starts too late may miss the sign preparation; a clip that ends too early may lose the final hold.
Example
- A 2-second clip at 30 frames per second has about 60 frames. A pose extractor may create one keypoint record per frame.
- For a sign with repeated movement, frames should show start position, movement, repeat, and final hold.
Practice
- Pick one sign or short phrase and sketch six key moments from beginning to end.
- Label the start, movement, hold, repeat if any, transition, and end.
Checkpoint
Before moving on
You can explain why timing is part of the sign, not just a video detail.
Pipeline note
Pipeline note
Keep frame counts and frame rate in metadata so training and review tools know how time was represented.