Skip to content

Implementation of the Automatic Emergency Braking System using deep learning.

License

Notifications You must be signed in to change notification settings

kuldeepbishnoi/Automatic-Emergency-Braking

Repository files navigation

Automatic Emergency Braking System

This repository contains code for a project whose goal is to Implement Automatic Emergency Braking System using a monocular camera. This model is trained on tusimple lane dataset.

Algorithm Used:

Brief Intro

LaneNet

LaneNet algorithm is a state of art deep convolution neural network which is used to detect lanes and is implemented using tensorflow.

LaneNet Output

DeepSORT

We use YOLO v3 algorithm to perform vehicle detections. Then we take this output feed it to DeepSORT in order to create a highly accurate vehicle tracker.

DeepSORT Output

Automatic Emergency Braking System

This feature can sense incoming(traffic coming to ego lanes) and slow(as well as stopped) traffic ahead and urgently apply the brakes.

Automatic Emergency Braking System Output

Installation

Required package could be installed by following the given steps.

  1. Download Github Repository.
  2. Download the weights from the given link and keep them in the similar folder structure as kept in the link.
  3. Install Anaconda.
  4. Run the given command in anaconda prompt.
conda env create -f requirements.txt

Test model

You can test the provided test frames on the model by following the given steps.

  1. Activate the environment in anaconda prompt by using the given command.
activate EBS
  1. Run the test_ebs file by using the given command.
python test_ebs.py