港大计算机系课程作业

Github 链接包含 COMP8501 Advanced topics in computer graphics 和 COMP9102 Data Management and Information Retrieval 课程作业。

https://github.com/louiswng/HKUCS-Coursework

COMP8501 Assignment1

Please accomplish one of the following two tasks:

1. Implement one of the traditional 2D Image-Based graphics algorithms:

   • Poisson Image Editing

   • Graph Cut

   • GrabCut

   You must reimplement the algorithm by yourself.

2. Implement one of the deep learning 2D Image-Based graphics algorithms:

   • Fully Convolutional Networks for Semantic Segmentation

   • CycleGAN

   You may use open source framework/implementation. But you have to train the model by yourself and include the trained model in your submission.

​ You may also choose other deep learning 2D Image-Based graphics algorithms from published paper or your own project.

​ The technical report should include detailed parameters showing the model convergence.

Requirements:

1. Technical report, source code, example images are expected to be submitted through Moodle.
2. There should be detailed instructions on how to run your code. TA will run your code against testing datasets and will deduct marks if the instructions are unclear.
3. The reimplemented algorithm is expected to have comparable performance with the original paper. You should evaluate the performance in the technical report.
4. Recommended programming languages: C++/Java/Python/MATLAB