Development

Start: 08/2020.

End: 12/2020.

Project description

YOLO (You Only Look Once) is a real-time object detection algorithm that can detect and identify objects in images and videos. It works by dividing an image into a grid of cells and predicting the object class and bounding box for each cell, using a single convolutional neural network. YOLO is known for its speed and accuracy, making it popular for various computer vision applications such as self-driving cars, surveillance, and robotics.

My undergraduate thesis consisted of detecting license plates for cars and pedestrians and comparing the performance of versions 3.0 tiny, 4.0 and 4.0 tiny. The code was developed using google colab to use some source of GPU. The train dataset was collected by a PhD student, but I used random street photos under different conditions to test the results. The results can be seen in the images below. I also recorded videos with YOLO performing in real-time.

Screenshot

Image 1
Comparison YOLO versions for license plates

Image 2
Comparison YOLO versions for license plates and pedestrians

Image 3
Comparison YOLO versions for license plates and pedestrians

Tools

  • Neural Networks
  • YOLO
  • Google Colab