Computer Vision
•
Identify objects on car and satelitte images
In this project, I managed to a UNet model using PyTorch and use this model to detect objects on images from the Carvana Image Masking Challenges and the Dubai Aerial Imagery datasets.
Image segmentation is a process of partitioning, or dividing, an image into multiple segments, thus called “image segmentation”. The goal of image segmentation is to identify objects by assigning different colors to different objects.
UNet is a CNN-derived Deep Learning architecture that consists of multiple encoders and decoders that are connecte by skip connections. As result, the entire architecture looks like a “U” shape.
I made two separates model, one model is used to segment car images, and the other one satellite images. With only 50 iterations, these models are capable of producing these results.
