This code shows the train and test of a YOLOV5 convolutional neural network for recognition of electronics components
see the model working on video: here
download the folder and upload to your google drive put in a dir named /Real_PCB/Databset/ :
here it's the dataset originaly created by Animeshkumar Nayak(available on Kaggle) with some improvements that i made in roboflow
This is how should look the projet struture :
MyDrive/
└── Real_PCB/
├── Dataset/
│ └── pcb_data/
│ ├── data.yaml
│ ├── test/
│ ├── valid/
│ └── train/
│
└── yolov5/
I chose to train on colab cause i havent a powerfull GPU as colab invidia does and its free :)
The train process is all commented
To test the models you only need the results. its also commented
yolov5 allows you to test the model using videos images even your web camera if you setup it your machine
(see more results here: here )