Related Course Projects
Image and Video Processing:
- Developed a neural image colorization model using ResNet-18 with perceptual and pixel-wise loss for grayscale-to-RGB translation.
- Implemented adaptive image segmentation using histogram clustering and region growing.
- Compared Mean Shift and Graph Cut segmentation methods across diverse image types.
- Applied classical CV techniques: template matching, thresholding (Otsu, Niblack), and histogram equalization.
- Built feature alignment pipelines with Harris corner detection, SIFT descriptors, and panorama stitching using RANSAC.
- Evaluated image denoising methods (Gaussian and average filters) using PSNR under varied noise levels.
- Constructed Gaussian and Laplacian pyramids for multi-scale image decomposition and reconstruction.
- Explored color and spatial processing by analyzing RGB/HSV color channels and applying 2D convolutions with various filters.
- Developed a block-based hybrid video coder using integer-accuracy EBMA for P-frame compression with intra/inter prediction.