DingDone: Autonomous Flight Drone for Searching Survivor with Monocular Depth Estimation and Object Detection

Requirments

Avoidance, Localization

• This code is tested with Keras 2.2.4, Tensorflow 1.13, CUDA 10.0, on a machine with an NVIDIA Titan V and 16GB+ RAM running on Windows 10 or Ubuntu 16.
• Other packages needed keras pillow matplotlib scikit-learn scikit-image opencv-python pydot and GraphViz for the model graph visualization and PyGLM PySide2 pyopengl for the GUI demo.
• Minimum hardware tested on for inference NVIDIA GeForce 940MX (laptop) / NVIDIA GeForce GTX 950 (desktop).
• Training takes about 24 hours on a single NVIDIA TITAN RTX with batch size 8.

Project Overview

1. avoiding obstacle with monocular depth estimation
2. detecting survivor with object detecting
3. estimating distance and direction of survivor from drone

Run

1. Clone this repository

$ git clone https://github.com/Ewha-BanBanBank/DingDone_final.git

2. Connect drone to your computer
3. Run Demo - avoiding obstacle

$ cd Avoidance
$ python3 DingDone_path.py

4. Run Demo - detecting survivor, estimating position of survivor and showing those information to application

$ cd ../Localization
$ python3 localization.py

References

1. Monocular Depth Estimation : https://github.com/ialhashim/DenseDepth
2. Object Detection : https://github.com/ultralytics/yolov5

Poster

Video