Computer-Vision-Assignments

Reports of all the assignments are available in ./Report directory.

Assignment-1

• Average Filter (3x3, 5x5, 11x11, and 15x15). Analysis of using avg filter with different kernel sizes.
• Adding Salt and Pepper noise. Removing noise using median filter with different kernel sizes.
• Analysis of using Gaussian kernel with differnent kernel sizes for Blur effect.
• Generated a Gaussian and Laplacian pyramid.
• Performed Discrete Wavelet Transform using Haar classifier to remove High frequency component to smoothen the images.
• Performed Watermarking using DWT.

Assignment-2

• Implemented Hough Tranform for circles from scratch.
• Camera Calibration using checkerboard.
• Implemented Harris Corner Detection from scratch.

Assignment-3

• Image segmentation using K-Means clustering for:
  • 3-D Color Space
  • 5-D space, using extra two dimensions corresponding to x and y co-ordinate of pixel.
• Image segmentation using Mean-shift clustering.
• Face detection in an images using:-
  • Skin color thresholding - Considered RGBA, HSV, YCrCb color space.
  • Seeded Segmentation - Manually marking point in an image and recursively analyzing nearby points.
• Image Classification using Bag-of-Visual-Word with HOG and LBP features on CIFAR-10 Dataset.

Assignment-4

• Image Classification: Extracted features from pretrained AlexNet model and classification using Random Forest Classifier.
• Convolutional Neural Network: Created our own architecture for CNN and performed 4 class classfication from CIFAR-10 Dataset.
• Image Identification and Matching: Matched using Brute-Force feature matching technique and plotted box for matched object in images.
• Panaroma Stiching: Stiched 3 Taj Mahal Images into 1.

Assignment-5 [Bonus]

• Computed Depth Map in given 2 images.