A visual interface developed with python to support object detection tests and output visualization. This application uses the YOLO convolutional neural network as a base object detector.
The first thing you should do is clone this repository to your computer. To this run the following command in your terminal.
user@computer:~$ git clone https://github.com/eluire/Yolo_app.gitFirst, you need to configure your system to run YOLO CNN, to this, visit the following repository.Click here https://github.com/vanluwin/enviroment)
After this, you are able to execute the YOLO CNN, in this repository, we provide a folder called darknet, open it and run the following command in your terminal
user@computer:~/yolo_app/darknet$ makeIf you want to know more about how the YOLO Convolutional neural network works, visit these sites.Official repository, Official site
We make available to you a specific weight file to vehicles detection, if you want to download this file click on the following link and save in this path: Yolo_app/darknet/weights .Click here
If you want to know how we create this weight file, see all of the statistics,detection metrics and more about our job with image processor and object detection, check the following link.[Click here](link para artigo)
But if you have your own file of weights just move to the darknet folder.
We make available to you some demo videos to detection,if you want to download this file click on the following link and save in this path: Yolo_app/darknet/videos .Click here
To install anaconda package manager follow the official installation guide.Click here
We make available a file named pyqt.yaml, which is no more than a complete setup with all needed packages to run this app.
Run the following command in your console:
user@computer:~/yolo_app$ conda env create -f pyqt.yamlIf your environment was created correctly, execute this to enter activate it:
user@computer:~/yolo_app$ conda activate pyqtSo now let's get started with Yolo app!
Inside of your virtual environment run the following command:
[pyqt]user@computer:~/yolo_app$ python yolo.pyif everything is correct, you will see this screen.
Now you should choose some essential files. If you already have your setup to test feel free to select other files. In this repository we make available default files to your test, just follow the steps below to set your test setup.
1. Select the video file.
Obs: this is a video that you want to detect.
2. Select the weight file.
3. Select the data file.
Obs: this is file contains information about the classes to detection.
4. Select the cfg file.
Obs: this is file contains a convolutional neural network configuration.
If everything is correct, you can star the detection. When the detection was finished the Yolo app will save a video file called 'video_detected'.
Now you can play your video already detected just select it and press play video stream button.