A selected (perfect score, or a score close to it) collection of projects from my undergraduate computer science courses, including computer vision, computer architecture, deep learning, etc.
Computer Architecture [link]
- using c++
- Project 1: Simple MIPS assembler
- Project 2: Building a Simple MIPS Emulator
- Project 3: Simulating Pipelined Execution
- Project 4: Multi-level Cache Model and Performance Analysis
Computer Vision [link]
- using python
- Project 1
- Reduce noise and enhance image contrast through filtering in both the spatial and frequency domains.
- Project 2
- Image classification using conventional analytical (mathematical) methods such as the Bayes classifier and a simple CNN.
- Project 3
- Perform camera calibration using a (cellphone) camera and implement a virtual ruler.
System Programming [link]
- using c, c++
- HW 1
- Implementation of simple bitwise operations
- HW 2
- Modify the provided code and implement an optimized procedure that performs the same functionality with improved performance.
- As a result, achieved a 4x performance improvement
- Team Project
- Programming involving robot control, communication with servers, path planning, strategy, and real-time QR code detection and decoding.
Digital Image Processing [link]
- using MATLAB
- HW 1
- Implementation of an interpolation method
- HW 3
- Implementation of noise filtering and edge enhancement in the spatial and frequency domains.
Deep Learning [link]
-
using python
-
Programming Assignment 1 : Neural Network design
- (Main) 3-layer Neural Network for Classification without the deep learning framework (only python)
- (Extra credit) 3-layer Neural Network for Classification using a deep learning framework (e.g. pytorch, tensorflow)
-
Programming Assignment 2 : Convolutional Neural Network
- (Main) Convolution Neural Network for Classification without deep learning framework
- (Extra credit) Visualize the activation map using Class Activation Map using the author provided code
-
Programming Assignment 3 : Recurrent Neural Network
- (Main) Recurrent Neural Network for sentiment analysis generation without deep learning framework
- If you cannot design the networks without DL framework, you can use DL framework (but, you will only get 30% score of total)