Multi-Category-Semantic-Segmentation-on-CT-scans-Using-Deep-Neural-Networks

Author: BAIM Mohamed Jalal
Student ID: 313551810
Date: November 20, 2024

---

Overview

This repository contains implementations of various deep learning models for medical image processing tasks. The models are designed to perform image segmentation on a given dataset. The following notebooks are included:

1. Deeplabv3Model.ipynb: Implements the DeepLabv3 architecture for image segmentation.
2. UnetplusplusModel.ipynb: Implements the UNet++ architecture for image segmentation.
3. SelfMadeModel.ipynb: Introduces a custom architecture proposed by the author, applied to the dataset.

---

Proposed Model

Repository Structure

├──codePackage ├── Deeplabv3Model.ipynb ├── UnetplusplusModel.ipynb ├── SelfMadeModel.ipynb ├── README.md

Update Dataset Paths

In each notebook, locate the Paths section (typically the second cell) and update the paths to point to your local dataset directories.

# Paths
train_img_dir = "/path/to/your/dataset/train_imgs"
train_label_dir = "/path/to/your/dataset/train_lbs"
test_img_dir = "/path/to/your/dataset/test_imgs"