https://github.com/abhineet123/deep_mdp

Deep MDP: A Modular Framework for Multi-Object Tracking
https://github.com/abhineet123/deep_mdp

deep-learning deep-neural-networks markov-decision-processes multi-object-tracking object-tracker object-tracking siamese-network siamese-neural-network siamese-tracker tracking

Last synced: 9 days ago
JSON representation

Deep MDP: A Modular Framework for Multi-Object Tracking

• Host: GitHub
• URL: https://github.com/abhineet123/deep_mdp
• Owner: abhineet123
• Created: 2022-04-21T18:58:36.000Z (almost 3 years ago)
• Default Branch: master
• Last Pushed: 2022-05-19T00:55:45.000Z (over 2 years ago)
• Last Synced: 2024-11-21T00:53:59.036Z (2 months ago)
• Topics: deep-learning, deep-neural-networks, markov-decision-processes, multi-object-tracking, object-tracker, object-tracking, siamese-network, siamese-neural-network, siamese-tracker, tracking
• Language: Python
• Homepage: https://arxiv.org/abs/2310.14294
• Size: 5.49 MB
• Stars: 2
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • Changelog: history.py

Awesome Lists containing this project

README

        


Welcome to the home of **Deep MDP**  - a fast, modular and parallelized Python implementation of the [MDP framework](https://github.com/yuxng/MDP_Tracking) for Multi-Object Tracking with added support for deep learning. 

- [Python dependencies](#python_dependencies_)

- [Installation in Linux](#installation_in_linu_x_)

    - [imbalanced-dataset-sampler](#imbalanced_dataset_sample_r_)

    - [libsvm](#libsv_m_)

    - [C_modules](#c_modules_)

- [Installation in Windows](#installation_in_window_s_)



# Python dependencies

```

pip3 install xlwt

```



# Installation in Linux



## imbalanced-dataset-sampler 

```

cd models/thirdparty/imbalanced-dataset-sampler && python3 setup.py install

```



## libsvm

 * install [libsvm](https://www.csie.ntu.edu.tw/~cjlin/libsvm/):

```

cd models/thirdparty/libsvm-3.23/python

make

cp ../libsvm.so.2 /usr/local/lib

mkdir ../../../libsvm

mkdir ../../../libsvm/linux

cp ../libsvm.so.2 ../../../libsvm/linux

mkdir ../../../libsvm

mkdir ../../../libsvm/linux

cp ../libsvm.so.2 ../../../libsvm/linux/libsvm.so.2

cd -

```



## C_modules

```

cd cmodules && mkdir build && cd build

 

cmake ..

 

sudo make install

```



# Installation in Windows

 * install [libsvm](https://www.csie.ntu.edu.tw/~cjlin/libsvm/) by following the instructions [here](https://stackoverflow.com/a/32358275) and [here](https://stackoverflow.com/a/12877497)

 * install [cmake](https://cmake.org/)

* install Visual Studio 2015 by following instructions [here](https://stackoverflow.com/a/44290942)

* install [OpenCV](https://opencv.org/) by following the instructions [here](https://docs.opencv.org/3.2.0/d3/d52/tutorial_windows_install.html)

* installation using pre built binaries is not guaranteed to work so build from source if cmake does not find OpenCV in the next step

* create an environment variable called `OpenCV_DIR` pointing to the OpenCV installation or build folder containing the `OpenCVConfig.cmake` file

* create folder `tracking_module/cmodules/build` and run `cmake ..` from there

* Open `MDP.sln` in Visual Studio, change build type to `Release` from `Debug`, right click on `All Build` in the Solution Explorer and select `Build`

* Once this completes, right click on `INSTALL` in the Solution Explorer and select `Build`