Project

This project can be used to obtain skeleton-data from openpose application. The following features are available:

• Write json from videos

• Create database from json files

• Read data from database

Hadware Configuration

• OS: Linux [Ubuntu 16.04 LTS - 64 bits]

• GPU: NVIDIA GeForce GTX 1060 (6GB)

• Memory: 16 GB

• Processor: Intel Core i7 (7th Gen)

Installation

1. Install openpose using this installation procedure.

2. Install virtualenv using step #3 from this post.

3. Download requirement.txt file in this link.

4. Create a new env using this command mkvirtualenv "name_env" -p python3 change "name_env" for your choice name.

5. Enter inside env: workon "name_env" .

6. Install dependences: pip -r requirements.txt.

7. Clone this respository using git clone https://github.com/ximenesfel/action_recognition_openpose.git command.

Usage

To write json files from videos (Recommended: GPU)

• Open write_json_from_video.py and fill this variables: path_videos (path to your videos folder) and path_output_data (path where you like to save json data). Please insert all path. Ex: /home/"user"/Desktop/.....

• Run python write_json_from_videos.py.

Create database from json files (Recommended: GPU or CPU)

• Open generate_data.py and fill the variable path_data (path where you save json files). Please insert all path. Ex: /home/"user"/Desktop/.....

• Run python generate_data.py.

• The file database.dat will be saved in your current directory with all data from json's files.

Read data form database (Recommended: GPU or CPU)

• Open read_data.py and fill the variable data_path (path where you save database file).

• Run python read_data.py.