Welcome to DroneVis’s documentation!

DroneVis is a full compatible drone library to automate computer vision algorithms on parrot drones. The library and computer vision alogirthms were built with real-time constraints on-mind.

All of the implemented real-time data, inference, and detection achieve a minimum fps >= 4.5 on an Intel core 8 CPU.

Github repository: https://github.com/ahmedheakl/drone-vis

Check out the Installation section for further information, including how to install the project.

Main Features

• Unified state-of-the art computer vision algoritms

• Full control over the drone

• PEP8 compliant (unified code style)

• Documented functions and classes

• Tests, high code coverage and type hints

• Clean code

• Multiple implementations for the same models

Drone Control	Detection Models	Human Counting
Right, Left	Faster R-CNN	CSRNet
Up, Down	CenterNet	…
Forward, Back	YOLO	…
Takeoff, Land	Single Shot Detector	…
Reset, Emergency	…	…
Rotate Left/Right	…	…
Hover, Caliberate	…	…

Note

This project is under active development.

User Guide

• Installation
  • Prerequisites
  • Windows 10
  • Linux
  • Development version
  • Docker
• Getting Started
  • 1. Graphical User Interface
  • 1. Command-Line Interface

Connections

• Drone Connection
  • Main Structure
  • Command Connection
  • Video Connection
  • Navigation Data Connection
• Drone
  • Drone class
• Command Thread
  • Sending Commands
  • Command Class
• Video Thread
  • Video Specs
  • Command Class

Detection

• Detection Algorithms
  • Detection on Your Web Camera
  • Different Model Implementations
  • Abstract Models
• Single Shot Detector
  • SSD Torch
• YOLO
  • YOLOv5 Torch
• Faster R-CNN
  • Faster R-CNN Torch

Pose Estimation and Segmentation

• Pose Estimation and Human Segmentation
  • PoseSegEstimation

Face Detection

• Face Detection

User Interfaces

• Graphical User Interface
• Command-Line Interface

Indices and tables

• Index

• Module Index

• Search Page