Skip to content

Latest commit

 

History

History
31 lines (25 loc) · 1.51 KB

sources.md

File metadata and controls

31 lines (25 loc) · 1.51 KB

Datasets:

Reading:

Project Idea: Detecting objects in artwork with R-CNN, Fast R-CNN, and YOLO/YOLOv2 (in PyTorch?)

  1. Start with pre-trained model and report results on PeopleArt, PhotoArt50
  2. Try training on PhotoArt/PeopleArt.
  3. Try "removing" style with neural style transfer from painting to photograph.
  4. Try "removing" style by tuning parameters in neural style transfer.

Presentation Goal:

  • Have pre-trained results of some R-CNN variant and YOLO on PeopleArt and PhotoArt50
  • (Hopefully) train one of R-CNN variant and YOLO on PeopleArt or PhotoArt50
  • Have basic neural style removal

Current progress:

  • Have some neural style removal, not good in general, but have one figure and can explain what we've tried and what we can keep trying
  • Have DataLoader for PeopleArt, PeopleArt photos
  • (David) Working on YOLO pre-trained - Pytorch version is really finicky... maybe I'll try the tensorflow version...