Skip to content

Latest commit

 

History

History
55 lines (40 loc) · 1.45 KB

README.md

File metadata and controls

55 lines (40 loc) · 1.45 KB
Raw

Bhojpur Vision - Processing Engine

The Bhojpur Vision is an advanced visual data processing engine applied within the Bhojpur.NET Platform for delivery of intelligent applications or services.

Build Source Code

You need to install Go v1.16 at least. Also, CUDA is required for faster GPU processing.

Object Detection Models

Firstly, you need to do the following to achieve automatic object detection using ML. The PASCAL VOC (in .xml file format) and DarkNet (in .txt file format) could be used for annotations.

  • Data Acquisition
  • Data Preparation according to the Yolo
  • Load Dataset
  • Train the dataset
  • Obtain the model weights
  • Test the model

Alternatively, you can use Yolo v5 pre-trained models. Please run the following commands to download sample object detection models.

./getModels.sh

Simple Examples

go run ./internal/image/main.go
go run ./internal/webcam/main.go

Model Annotation Application

Firstly, install the Qt5 and libxml2 dependencies

cd label
brew install qt
brew install libxml2

then, you can run the following commands to launch the Model Annotation application

pip3 install -r requirements.txt
make qt5py3
python3 visionlab.py