al-shafi-github / blood_vessel_segmentation_unet

Blood Vessel Segmentation of Diabetic Retinopathy Fundus Image

Blood_Vessel_Segmentation Using Unet Architecture

Blood Vessel Segmentation of Diabetic Retinopathy Fundus Image. The dataset on the Unet model trained are Drive2004 Dataset and Chasedb

Here is the segmented image picture

UNet is a convolutional neural network (CNN) architecture. The architecture of UNet is characterized by its unique U-shaped design, which consists of an encoding path and a decoding path. The encoding path involves a series of convolutional and pooling layers that progressively reduce the spatial dimensions of the input image while capturing essential features. The decoding path, on the other hand, involves upsampling and concatenation operations to restore the spatial resolution and produce the final segmentation map. UNet is known for its ability to effectively capture both local and global context, making it well-suited for segmentation tasks where precise boundary delineation is required. Its skip connections between the encoding and decoding paths enable the propagation of feature information, allowing the model to maintain high-resolution details throughout the process.

Unet architecture: