soumilchugh / LaneDetection

This repo contains tensorflow implementation of UNET architecture for semantic segmentation of lanes

Repository from Github https://github.comsoumilchugh/LaneDetection

autoencoder autonomous-vehicles cnn computer-vision lane-detection python segmentation semantic-segmentation stochastic-gradient-descent tensorflow tensorflow-models unet

LaneDetection

Dataset - BDD100K

Usage

Generate labels for each of the trained images using Canny edge detector
Train a semantic segmentation model on these train images

Examples of Raw images

Examples of Labels for the Segmentation model

Labels superimposed on the original images

Output from the Trained Model

Output Superimposed on the images

Feature Map Visualisation of some of the layers of the network

About

This repo contains tensorflow implementation of UNET architecture for semantic segmentation of lanes

autoencoder autonomous-vehicles cnn computer-vision lane-detection python segmentation semantic-segmentation stochastic-gradient-descent tensorflow tensorflow-models unet

Languages

Language:Python 100.0%