Skip to content

Latest commit

 

History

History
22 lines (18 loc) · 800 Bytes

README.md

File metadata and controls

22 lines (18 loc) · 800 Bytes

Detection benchmarks

The benchmarks/detection folder contains the code for the following models:

  • Faster R-CNN
  • DASS
  • YOLOv8
  • Grounding Dino
  • Magi
  • SSD (to be added)

The benchmarks are run on the validation split of the CoMix dataset, which comprehends:

  • Comics-style:
    • comics100 (val, test, dev[todo])
    • DCM (val, test)
    • eBDtheque (val, test)
  • Manga-style:
    • PopManga (val, test)
    • Manga109[todo] (val, test, dev)

For these datasets, follow the instructions in the CoMix/README.md and then datasets/README.md files to preprocess and convert the data.

Here you can find the instructions for running the models generation and evaluating the predictions evaluation.