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.
- LaneNet
- DeepSORT (YOLO v3)
- Automatic Emergency Braking System
LaneNet algorithm is a state of art deep convolution neural network which is used to detect lanes and is implemented using tensorflow.
LaneNet Output
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
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
Required package could be installed by following the given steps.
- Download Github Repository.
- Download the weights from the given link and keep them in the similar folder structure as kept in the link.
- Install Anaconda.
- Run the given command in anaconda prompt.
conda env create -f requirements.txt
You can test the provided test frames on the model by following the given steps.
- Activate the environment in anaconda prompt by using the given command.
activate EBS
- Run the test_ebs file by using the given command.
python test_ebs.py