RAPID VISION

Vehicle Gate Processing System powered by advanced computer vision. Detect plates, count passengers, analyze demographics, and track individuals.

MODEL DEMOS

LPR DETECTION

License Plate Recognition using YOLOv9 detection and PaddleOCR text extraction. Optimized for Tanzanian plate formats.

Speed: 15-30ms

Accuracy: ~95%

HEAD DETECTION

YOLOv8 model trained on CrowdHuman dataset for accurate head detection in crowded vehicle scenes.

Speed: ~25ms

Model: YOLOv8m

FACE ANALYSIS

DeepFace-powered age estimation and demographic classification. RetinaFace backend for robust detection.

Speed: 50-100ms

Features: Age, Race

PERSON RE-ID

OSNet model for cross-camera person matching. Generates 512-dimensional embeddings for identity tracking.

Speed: ~5ms

Embedding: 512-dim

FULL PIPELINE

Run the complete vehicle processing pipeline. Combines LPR detection, head counting, face analysis, and Re-ID for comprehensive vehicle entry processing.

Total Time: 200-350ms

Steps: 4 Models

Output: JSON Payload

RAPID v1.0.0 · SkyConnect · Ngorongoro Conservation Area