Terra Object Detection (v. 0.9.0 Beta)

An application based on ImageAI library which can run object detection tasks from device cameras and IP cameras in real time.

• Detects and saves detections as separate images as well as immediately notifies user about detection
• Availability to choose detection speed and accuracy
• Up to 8 cameras can be saved and work simultaneously in the current version

Object Detection

An application uses YOLOv3 and TinyYOLOv3 models for object detection.

Requirements for Developers

• Python 3.7.4
• ImageAI 2.1.5
• Tensorflow 1.11
• OpenCV 4.1.0.25
• Keras 2.2.4
• yolo.h5 and yolo-tiny.h5 models

Installation

1. Install ImageAI 2.1.5 and all necessary dependencies

• ImageAI:

pip3 install imageai --upgrade

• Tensorflow:

pip3 install tensorflow==1.11.0

• OpenCV:

pip3 install opencv-python

• Keras:

pip3 install keras

2. Download yolo.h5 and yolo-tiny.h5 and locate them together with other files in venv\ directory

• Yolo.h5
• Yolo-tiny.h5

3. Locate __ init __.py in venv\Lib\site-packages\imageai\Detection (change the existing file)

Run _appearance.py to launch the application.

License

This project is licensed under the MIT License.

Contacts

• avrilo@tutanota.com