You can deploy this model and start earning money today!

bengaliAI/
badlad-mrcnn-paper

Name: badlad-mrcnn-paper
Availability: InStock

detectron2

document-layout-analysis

badlad

mask-rcnn

instance-segmentation

Views

No views yet

BaDLAD Mask R-CNN — Paper Baseline

Mask R-CNN R-50-FPN trained on the BaDLAD training set for Bengali document layout analysis (instance segmentation track). Matches the paper’s Mask R-CNN / ImageNet setup used for mask mAP reporting.

Paper: BaDLAD (ICDAR 2023)

Code: BengaliAI/BADLAD

Dataset: BaDLAD on Kaggle

Project page: bengaliai.github.io/badlad

Model Details

Field	Value
Architecture	Mask R-CNN R-50-FPN 3x
Framework	Detectron2
Config	`COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml`
Classes	paragraph, text_box, image, table (4)
Training data	BaDLAD train (~20 365 images, 6 domains)
Iterations	10 000
Init	ImageNet
Checkpoint	`model_final.pth` (~169 MB)
sha256	`d3b663446a3aeecfbf1d5f437110257948a2616eec5496504f9306d65460f497`

Usage

python

from detectron2 import model_zoo
from detectron2.config import get_cfg
from detectron2.engine import DefaultPredictor
from huggingface_hub import hf_hub_download

weights = hf_hub_download("bengaliAI/badlad-mrcnn-paper", "model_final.pth")

cfg = get_cfg()
cfg.merge_from_file(
    model_zoo.get_config_file("COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml")
)
cfg.MODEL.ROI_HEADS.NUM_CLASSES = 4
cfg.MODEL.WEIGHTS = weights
cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.05
predictor = DefaultPredictor(cfg)

Evaluation

Re-run on the paper hidden test (13 328 images). Primary metric: COCO mask AP@[.5:.95], score_thresh=0.05.

Metric	Value
mask_mAP	0.331
mask_mAP50	0.558
bbox_mAP	0.343
bbox_mAP50	0.562

Per-class mask AP: paragraph 0.609, text_box 0.231, image 0.411, table 0.075.

Domain-wise mask AP (×100) — this checkpoint / paper Table 3 (M-RCNN | ImgNet | Mask):

Domain	n	P	Tx	I	Tb
Historical Newspapers	345	60.3 / 60.3	18.3 / 18.3	57.3 / 57.3	0.0 / 0.0
New Newspapers	65	41.4 / 41.4	13.1 / 13.2	45.2 / 45.2	1.9 / 1.9
Magazine and Books	11674	61.8 / 61.8	25.3 / 25.3	44.9 / 44.9	2.3 / 2.3
Liberation War Documents	402	71.1 / 71.2	26.8 / 26.8	1.1 / 1.0	40.1 / 40.1
Government Documents	514	49.4 / 39.1	23.7 / 18.7	26.1 / 19.4	5.1 / 3.7
Property Deeds	328	38.0 / 0.6	14.2 / 0.7	13.3 / 2.1	3.2 / 0.6

Related models

Repo	Role
`bengaliAI/badlad-frcnn-paper`	Faster R-CNN (bbox)
`bengaliAI/badlad-yolov8m-seg`	YOLOv8m-seg

Citation

bibtex

@inproceedings{shihab2023badlad,
    title     = {{BaDLAD}: A Large Multi-Domain {Bengali} Document Layout Analysis Dataset},
    author    = {Shihab, Md. Istiak Hossain and Hasan, Md. Rakibul and Emon, Mahfuzur Rahman and Hossen, Syed Mobassir and Ansary, Md. Nazmuddoha and Ahmed, Intesur and Rakib, Fazle Rabbi and Dhruvo, Shahriar Elahi and Dip, Souhardya Saha and Pavel, Akib Hasan and Meghla, Marsia Haque and Haque, Md. Rezwanul and Chowdhury, Sayma Sultana and Sadeque, Farig and Reasat, Tahsin and Humayun, Ahmed Imtiaz and Sushmit, Asif Shahriyar},
    booktitle = {Proceedings of the 17th International Conference on Document Analysis and Recognition (ICDAR)},
    year      = {2023},
    url       = {https://arxiv.org/abs/2303.05325},
}