Diskussion:Signalverarbeitung mit MATLAB und Künstlicher Intelligenz

#	date	plan for the next week	work progress
1	14.11.2024	safety briefing for the lab SVN/Wiki login time report in Sciebo make a mind map of your project and discuss it with Prof. Schneider	Done with mind map and shared with Prof. Schneider Got Wiki Author acess from Mr Robin Schröde
2	22.11.2024	mind map givs an overview intensify the research project plan and documentation as a gantt-chart discuss research results and project plan with Prof. Schneider	Researched on Neurons and how to create an artificial neuron network and Mathlab AI toolkits Included mindmap on project Page as a sub topic Gathering research results to discuss with Prof. Schneider alongside designing of the proejct plan
3	06.12.2024	What tasks of an image processing can be done by AI? E.g. Lane Detection Can AI be used to easily set up controllers like Line-Follower? Can AI be used to easily detect objects? Let us identify tasks that are solved in a classic way and with AI. What can the MATLAB Toolboxes be used for?	Researched on image processing Researched on how AI image processing works Looked at various image processing task that can be done by AI Worked on circular object detection with Houghs Transform Worked on person, dog and cat detection with YOLO models
4	13.12.24	Research on methods of doing lane tracking with AI	Researched on methods that lane tracking can be achieved with AI Installed the Reinforcement learning toolbox Looked at how the reinforcement toolbox works
5	20.12.24	Do lane keeping using CNN	Got data from Evra Trained the CNN on data collected Saved the simulation output as video alongside the training curve
6	27.12.24	Traffic sign detection and Number of M&M detections in an image	Found a database on Kaggle for Traffic signs which had more than 800 images, but it wasnt accurate as i couldnt detect all the signs in the test image Looked for another database and used it Research on image labeling softwares for AI Found an image labeling software and using it for labeling applications of the Yolo model
7	02.01.2025	Meet Prof. Dr.-Ing. Schneider and present work progress	Meet Prof. Dr.-Ing. Schneider and got corrections alongside recommendations Created a pre -trained network with CNN and saved the output Researched on more methods of achieving lane keeping with AI Got more test data from Evra Working on article to post on Wiki alongside research on the topic Learnt how to use school SVN better with Prof. Dr.-Ing. Schneider guidance
8	02.01.2025	Results of the training ar not usen use trained network to improve this make a MATLAB-Script that uses the trained NN to drive and produce results research methods for lane tracking with AI write down the results and sources weigh the solutions with a Morphological analysis choose the best solution implement it test it dokument everything in Wiki	Task 1 not finished input data is wrong, no left curve 100% right is too much, I would expect a steering in between -20%-20%
9	19.01.2025	Research on methods for lane tracking with AI	Transfer Learning Training a CNN from Scratch Transfer Learning (ResNet-50, MobileNetV3) + PID/MPC Control Reinforcement Learning
10	26.01.2025	Training a CNN from Scratch	Got more Data from Evra Edited the code to convert all the CSV files to .mat for training Wrote a code to train the network and normalized it to range from 0.2 to -0.2
11	09.02.2025	Training a CNN from Scratch	Got more Data from Evra Got Data for Object Detection Continued Methods to achieve Better training
12	16.02.2025	Training a CNN using Transfer Learning	Used Training Data Fine-tuned it with ResNet50 Saved the trained network
13	28.02.2025	Training a CNN using transfer learning but not using the .mat files	Fine-tuned it with ResNet50 Used Hough Transforms to detect the lines Saved Trained network
14	07.03.2025	Using YOlo pretrained network for Visual Relationship Detection with AI	Identify Objects Create Linkage Test on Image
15	13.03.2025	Researching on Signal Processing Trying to complete other Task Looking for ways to improve Traing for already done Methods of lane keeping	Training a New CNN network from Scratch Labeling data for object Detection Doing other task