Projects

Work I have built, led, and contributed to, plus earlier student projects.

Selected work

2 projects
  1. ADS-Net: Single Image Dehazing

    A hierarchical architecture bridging content-aware spatial processing with global frequency enhancement for image dehazing. Features an Adaptive Dynamic Feature Split Unit, a Dual-Stage Frequency-Aware Local Fusion module, and a Spatio-Fourier Global Fusion module. Achieves state-of-the-art performance on RESIDE, Dense-Haze, NH-HAZE, and O-HAZE benchmarks, with 42.32 dB PSNR on RESIDE-Indoor.

    PyTorchComputer VisionImage RestorationFrequency Domain
    No public link
  2. Journal Insight

    An intelligent assistant for academic paper analysis and summarization — an end-to-end AI demo that bridges research and usability.

    LLMPythonRAG
    No public link

Archive

1 project

Student projects, hackathons, and experiments, kept for the record.