🚗 Traffic Scene Semantic Segmentation

Custom AlexNet-based Encoder–Decoder Architecture

Semantic segmentation of traffic scenes using PyTorch and the BDD100K dataset.

Upload a traffic scene image and click Predict Segmentation.

📤 Upload Traffic Image

🌈 Overlay Result

🎯 Segmentation Mask

Status

📊 Model Information

Model: Custom AlexNet Encoder–Decoder

Dataset: BDD100K

Framework: PyTorch

Semantic Classes: 19

Device: CPU

📈 Performance

✅ Validation Loss : 0.7282

✅ Pixel Accuracy : 77.90%

✅ Mean IoU : 0.2564

📖 About

This application performs pixel-wise semantic segmentation of traffic scene images.

Each pixel is assigned to one of 19 semantic classes, allowing the model to identify roads, vehicles, buildings, vegetation, sky, pedestrians, and other important traffic scene objects.

The model is built using a Custom AlexNet-based Encoder–Decoder Architecture and trained on the BDD100K dataset.

🎨 Semantic Class Color Legend

	Road
	Sidewalk
	Building
	Wall
	Fence
	Pole
	Traffic Light
	Traffic Sign
	Vegetation
	Terrain
	Sky
	Person
	Rider
	Car
	Truck
	Bus
	Train
	Motorcycle
	Bicycle

👨‍💻 Developed by

Sandeep

🔗 GitHub Repository:

https://github.com/Sandeepkumarreddy-7/Traffic-Segmentation-AlexNet

⭐ If you like this project, consider giving it a star on GitHub!