mende237/pedestrian-tracking-yaounde

# Pedestrian Tracking & Trajectory Analysis — Carrefour Melen, Yaounde, Cameroon > **"Pedestrian Trajectory Extraction in Unstructured Urban Traffic: A Case Study from Yaounde, Cameroon"** A computer-vision pipeline that **detects, tracks, and analyses pedestrian trajectories** from video footage filmed at **Carrefour Melen, Yaounde**, and compares the observed motion patterns against the ETH/UCY benchmark datasets used in state-of-the-art trajectory prediction research (Social Force, Social-Transmotion, etc.). --- ## Motivation Pedestrian trajectory datasets from Low- and Middle-Income Countries (LMICs) are almost entirely absent from the literature. This project bridges that gap by: 1. Applying a **YOLOv8 + ByteTrack** pipeline to unstructured urban traffic footage from **Carrefour Melen, Yaounde**. 2. Extracting trajectories in the standard ETH/UCY CSV format (`id, frame, x, y`). 3. Quantitatively comparing speed distributions, path linearity, and pedestrian density with European datasets. --- ## Data Collection | Parameter | Value | |---|---| | **Location** | Carrefour Melen, Yaounde, Cameroon | | **Camera** | Samsung Galaxy A56 | | **Resolution** | Full HD — 1920 × 1080 px | | **Frame rate** | 30 FPS | | **Files** | `data/Record_1.mp4` (10m01s, 1277 MB), `data/Record_2.mp4` (5m00s, 639 MB) | | **Total footage** | ~15 minutes | > Video was recorded handheld / from a fixed vantage point at the intersection. > The 30 FPS rate provides sub-33 ms temporal resolution — sufficient for fine-grained speed estimation. ### Map **Coordinates:** 3°51'50.5"N 11°29'47.8"E (decimal: 3.864028, 11.496611) > Click the map to open interactively in OpenStreetMap. Also viewable on Google Maps. --- ## Results | Metric | Yaounde | ETH/UCY | |---|---|---| | Mean speed (m/s) | _TBD_ | ~1.0 | | Median linearity | _TBD_ | ~0.85 | | Density (ped/m2) | _TBD_ | _TBD_ | > Results will be updated once analysis is complete. --- ## Stack | Component | Tool | |---|---| …