Entry Task - Sentiment Analysis of Cephalopods (GSoC 2026)

Goal

The goal of this entry task is to assess an applicant’s basic readiness for a research-oriented project: data handling, simple feature extraction, reproducibility, and the ability to reason about behavioral signals. This task is intentionally small and exploratory.

Task Description

Data handling

• Select a short video clip (10–30 seconds) of an aquatic animal (cephalopod preferred).
• Load the video using Python and report frame rate, resolution, and duration.

Feature Extraction

Implement any two simple, interpretable features (choose two):

• Frame-to-frame motion magnitude (frame differencing or optical flow)
• Area or extent of movement over time
• Color or intensity histogram changes across frames
• Approximate contour size or bounding-area variation

Visualization

Plot the extracted features over time and include at least one visualization that plausibly supports behavioral interpretation (e.g., low vs high activity periods).

Short Analysis

Write a 5–10 sentence explanation describing what behavioral information the selected features may capture, key limitations, and how additional modalities (audio, posture) could improve inference.

Submission Format

Publish a public GitHub repository containing:

• Source code (Python script or Jupyter notebook)
• Any small sample data used, or clear instructions to obtain it
• A README with setup instructions, a short description of implemented features, and sample plots/screenshots

Evaluation

Evaluated on correct data handling, code clarity, quality of reasoning, and reproducibility. Advanced models or large datasets are not required.