AMMF: Attention-based Multi-phase Multi-task Fusion Network for Robust 3D Detection in Autonomous Driving

This repository contains the public release of the Python implementation of our AMMF network for 3D object detection.

[AMMF: Attention-based Multi-phase Multi-task Fusion Network for Robust 3D Detection in Autonomous Driving]

Bangquan Xie, [Zongming Yang], [Liang Yang], [Ruifa Luo], [Jun Lu], [Ailin Wei], [Xiaoxiong Weng] and [Bing Li]

Getting Started

Implemented and tested on Ubuntu 20.04 with Python 3.6 and Tensorflow-gpu 1.15.0 with CUDA 10.1, pytorch 1.7.1.

Clone this repo

git clone https://github.com/b-xie/AMMF.git --recurse-submodules

If you forget to clone the wavedata submodule:

git submodule update --init --recursive

Create environment

conda create -n ammf python=3.6 pip
conda activate ammf
pip install --upgrade pip

Install Python dependencies

cd AMMF/ammf
pip install -r requirements.txt
pip install tensorflow-gpu==1.15.0
pip install torch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2

Compile integral image library in wavedata

cd utils
sh scripts/install/build_integral_image_lib.bash

ammf uses Protobufs to configure model and training parameters. Before the framework can be used, the protos must be compiled (from top level ammf folder):

cd ../..
sh ammf/protos/run_protoc.sh

Alternatively, you can run the protoc command directly:

protoc ammf/protos/*.proto --python_out=.

Training

Dataset

To train on the Kitti Object Detection Dataset:

Download the data (left color images, Velodyne point clouds and other if you need) and place it in your home folder at ~/Kitti/object
Go here and download the train.txt, val.txt and trainval.txt splits into ~/Kitti/object. Also download the planes folder into ~/Kitti/object/training

The folder should look something like the following:

Kitti
    object
        testing
        training
            calib
            image_2
            label_2
            planes
            velodyne
        train.txt
        trainval.txt
        val.txt

Customized dataset:

https://drive.google.com/drive/folders/1A-_wfcO_BthlOlGONPXTSVSSX0DcqQPN?usp=sharing

Run Trainer

Public soon...

LICENSE

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Acknowledgement

Our implementation leverages on the source code from the following repositories:

https://github.com/kujason/avod