Diskussion:Image Processing with MATLAB and AI

#	Date	Planning for the week	Progress
1	01.01.2026	Plan bachelor thesis with Gantt-chart. Register to MATLAB Grader. Understand requirements and set up SVN.	Created Gantt chart (Timeline: Jan–Apr). Analyzed requirements and set up development environment.
2	08.01.2026	Solve Image Processing Tasks (Restoration & Enhancement). Implement Wiener Deconvolution and CLAHE.	Implemented Wiener Deconvolution (deconvwnr) to recover original images from degradation. Applied CLAHE (Contrast-Limited Adaptive Histogram Equalization) to improve visibility. Saved results to SVN.
3	15.01.2026	Solve Image Processing Tasks (Noise & Segmentation). Implement Median Filtering and K-Means.	Utilized Median Filtering (medfilt2) to remove salt-and-pepper noise. Applied K-Means Clustering (imsegkmeans) to partition images based on pixel values.
4	22.01.2026	Finalize Classic Algorithms (Model Segmentation & Classification). Document progress in Wiki.	Implemented Active Contours (Snakes) for specific object isolation. Developed Rule-Based Shape Analysis (regionprops) to classify objects (e.g., Square vs. Circle) without ML. Completed Task 1 (Classic Algorithms).
5	29.01.2026	Research on AI Solutions. Literature review on Deep Learning Architectures.	Conducted comparative analysis on DL Architectures: ResNet (Standard) MobileNetV2 (Edge Efficiency) EfficientNet (Scalability) Vision Transformers (ViT)
6	05.02.2026	Implement AI Image Processing (Classification). Design CNN architecture.	Designed a **Sequential CNN** with a convolutional base and dense classification layer. Successfully classified low-resolution color images into 10 distinct classes (airplanes, birds, etc.).
7	12.02.2026	Research Object Detection (YOLO). Data Collection and Labeling.	Selected YOLOv11 (Ultralytics) for the LEGO Parts Detection System. Collected custom dataset (~400 images). Labeled dataset (bounding boxes) for LEGO bricks.
8	19.02.2026	Train and Deploy YOLO Model. Run inference on test images.	Implemented the YOLOv11 Model using Google Colab and NVIDIA GPU acceleration. Trained the model on the custom LEGO dataset. Successfully ran inference: The system automatically identifies and classifies LEGO bricks in new images.
10	02.03.2026	Research Deep Learning Architectures Comparative Analysis of: ResNet (Standard) MobileNetV2 (Edge Efficiency) EfficientNet (Scalability) Vision Transformers (ViT)	Reviewed key literature for selected architectures. ResNet (Standard) MobileNetV2 (Edge Efficiency) EfficientNet (Scalability) Vision Transformers (ViT
11	09.03.2026	Identify key differences in: Model architecture Computational complexity Accuracy and performance Suitability for edge devices Prepare a comparison table for ResNet, MobileNetV2, EfficientNet, and Vision Transformers	To do