Portfolio project by @pkim02.
Implementation proof: see IMPLEMENTATION.md for the FCN-style dense prediction pipeline, upsampling weights, mIoU evaluation, and custom MySegModel variant.
A semantic segmentation project implementing FCN-style dense prediction and a custom segmentation model in PyTorch.
semantic_segmentation.ipynb: Semantic Segmentation with FCN-Style Models
- FCN-style conversion of classification features into dense prediction maps.
- Skip/upscaling style segmentation workflow with cross-entropy training.
- Validation using mean Intersection-over-Union style evaluation.
- A custom MySegModel variant using higher-resolution features and dilated context.
The executed notebook keeps training losses, segmentation visualizations, and the final discussion of the custom model versus baseline FCN behavior.
The notebooks keep most executed outputs so reviewers can inspect the results directly on GitHub. Full reproduction may require downloading the referenced public datasets or pretrained weights.
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
jupyter notebookOpen the notebook listed above and run cells in order. GPU is optional for review, but recommended for rerunning training-heavy experiments.
- Fully Convolutional Networks for Semantic Segmentation: https://arxiv.org/abs/1411.4038
- Multi-Scale Context Aggregation by Dilated Convolutions: https://arxiv.org/abs/1511.07122
- PASCAL VOC dataset: http://host.robots.ox.ac.uk/pascal/VOC/
This repository contains a cleaned portfolio version of my own implementation work. Assignment prompts, submission metadata, personal identifiers, and course-provided local figures were removed. The MIT license applies only to the code and documentation in this repository. Papers, datasets, pretrained weights, and any third-party libraries or assets keep their original licenses and terms.