small-models-for-glam/detr-resnet-50_nls-chapbooks

detr-resnet-50_nls-chapbooks

A DETR model fine-tuned to detect printed illustrations on chapbook pages, trained on the `biglam/nls_chapbook_illustrations` dataset (illustration bounding boxes from National Library of Scotland chapbook scans). Single-class output: early_printed_illustration.

For new projects, consider `small-models-for-glam/historic-newspaper-illustrations-yolov11` — a more recent YOLO-based illustration detector, faster at inference. This DETR model is kept for reproducibility and for the chapbook-specific use case it was trained on.

Usage

from transformers import pipeline

pipe = pipeline(
    "object-detection",
    model="small-models-for-glam/detr-resnet-50_nls-chapbooks",
)

pipe("https://huggingface.co/small-models-for-glam/detr-resnet-50_nls-chapbooks/resolve/main/Chapbook_Jack_the_Giant_Killer.jpg")
# [{'box': {'xmax': 290, 'xmin': 70, 'ymax': 510, 'ymin': 261},
#   'label': 'early_printed_illustration',
#   'score': 0.998}]

Training

Fine-tuned from `facebook/detr-resnet-50` for 10 epochs (lr=1e-4, batch_size=8, Adam, linear LR schedule).

Framework versions

• —Transformers 4.20.1
• —Pytorch 1.12.0+cu113
• —Datasets 2.3.2
• —Tokenizers 0.12.1

Example image credits

• —Wikimedia Commons — Chapbook_Jack_the_Giant_Killer.jpg
• —McGill Library — PN970_G6_V3_1846-1180

Part of the small-models-for-glam collection.