This is a demo of the Dash interactive Python framework developed by Plotly.
Dash abstracts away all of the technologies and protocols required to build an interactive web-based application and is a simple and effective way to bind a user interface around your Python code. To learn more check out our documentation.
Try out the demo app here.
To get started, select a footage you want to view, and choose the display mode (with or without bounding boxes). Then, you can start playing the video, and the visualization will be displayed depending on the current time.
First create a virtual environment with conda or venv inside a temp folder, then activate it.
virtualenv dash-object-detection
# Windows
dash-object-detection\Scripts\activate
# Or Linux
source venv/bin/activate
Clone the git repo, then install the requirements with pip
git clone https://github.com/plotly/dash-object-detection.git
cd dash-object-detection
pip install -r requirements.txt
Run the app
python app.py
The videos are displayed using a community-maintained Dash video component. It is made by two Plotly community contributors. You can find the source code here.
All videos used are open-sourced under Creative Commons. The original links can be found here.
The object detection model is the MobileNet v1, made by Google and trained on the COCO dataset. You can find their implementation on their official Github repo. You are encouraged to try this app with other models.
The data displayed in the app are pregenerated for demo purposes. To generate the csv files containing the objects detected for each frame, as well as the output video with bounding boxes, please refer to utils/generate_video_data.py. You will need the latest version of tensorflow and OpenCV, as well as the frozen graph ssd_mobilenet_v1_coco, that you can download in the Model Zoo. Make sure to place the frozen graph inside the same folder as generate_video_data.py, i.e. utils.
- Dash - Main server and interactive components
- Plotly Python - Used to create the interactive plots
- OpenCV - Create the video with bounding boxes
- Tensorflow - Generate the bounding box data
Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests to us.
- Xing Han Lu - Initial Work - @xhlulu
See also the list of contributors who participated in this project.
This project is licensed under the MIT License - see the LICENSE.md file for details
