Developed 6 algorithms in MATLAB based on an IEEE paper and verified each of their effectiveness in identifying shapes like squares, rectangles, triangles, circles, pentagons and diamonds from within images. The image to be analysed can be obtained as input from a webcam or by opening an image stored within the computer. Links: Binary Pattern Identification PDF, Binary Pattern Recognition Code, IEEE Paper PDF

Used Matlab to help identify license plate numbers. Using a standard license plate font, created standard templates for all the alphanumeric digits. Post several stages of image processing, the captured images were binarized and divided into separate regions of possible characters. The correlation between these regions and the standard templates were identified to obtain the plate characters and thus track the vehicle. Links: License Plate Recognition PDF, License Plate Recognition Excel, License Plate Recognition Code

Used pytorch in python to create a deep learning framework to classify the 10 type of objects in the CIFAR-10 dataset. Conducted experiments by changing the number of hidden layers, learning optimizers, batch normalization, adjusting learning rates and momentum to try to improve learning quality. After fine tuning the parameters, was able to reach an accuracy of 80.2%. Links: Python Code