DATASET
4,000+
Augmented Training Images
Computer Vision & Environmental AI
DATASET
Augmented Training Images
MODELS
YOLOv5x & YOLOv8m
DEPLOYMENT
Drone-Ready Architecture
PRECISION
Detection Precision
COST
Inspection Cost Reduction
When pollutants evade traditional enforcement
Urban health is compromised when vehicle emissions go unmonitored at scale. Relying on manual inspections creates an enforcement gap that allows high-polluting vehicles to remain undetected.
The Intellema Design Challenge
Air pollution from vehicle emissions poses significant public health risks, yet traditional manual inspection methods are too slow and costly for city-wide monitoring.
By training computer vision (CV) models on thousands of augmented images, the system provides a scalable solution for CCTV and drone networks. This architecture transforms passive surveillance into an active tool for environmental regulation.
High-accuracy YOLOv5x and YOLOv8m models trained on a dataset of over 4,000 augmented smoke-emission images.
Advanced resizing, normalization, and augmentation techniques to improve model robustness and generalization across diverse environments.
Comprehensive benchmarking of detection performance across multiple models and optimizers to balance accuracy and processing efficiency.
Modular architecture designed for seamless integration with CCTV and drone feeds to support real-time urban monitoring.
