AI Paper Atlas

Research map / Detection and segmentation / CNN features

Document Structure Extraction using Prior based High Resolution Hierarchical Semantic Segmentation

Mausoom Sarkar, Milan Aggarwal, Arneh Jain, Hiresh Gupta, Balaji Krishnamurthy

ECCV 2020 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation · 16 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add hierarchical high-resolution strip-wise segmentation (Highresnet) over low-res LowresnetLowresnet · Forms Dataset: ChoiceGroupMIoU (%): 72.61 -> 83.01Improved
add prior mask of previous strip when predicting next strip (ablation)Highresnet · Forms Dataset: ChoiceGroupMIoU (%): 80.14 -> 83.01Improved
add prior mask of previous strip when predicting next strip (ablation)Highresnet · Forms Dataset: ChoiceGroup TitleMIoU (%): 78.89 -> 80.87Improved
use 1D dilated convs instead of 2D dilated convs (ablation)Highresnet · Forms Dataset: ChoiceFieldMIoU (%): 81.95 -> 84.05Improved
use 1D dilated convs instead of 2D dilated convs (ablation)Highresnet · Forms Dataset: TextBlockMIoU (%): 89.71 -> 90.55Improved
predict all four hierarchies jointly instead of one model per hierarchy (ablation)Lowresnet · Forms Dataset: ChoiceGroupMIoU (%): 70.55 -> 72.61Improved
predict all four hierarchies jointly instead of one model per hierarchy (ablation)Lowresnet · Forms Dataset: ChoiceGroup TitleMIoU (%): 63.95 -> 69.03Improved
use a single shared final layer for all hierarchies instead of separate decoders (ablation)Lowresnet · Forms Dataset: ChoiceGroupMIoU (%): 69.37 -> 72.61Improved
use Highresnet instead of ImageNet-pretrained DeepLabV3+DLV3+ ImageNet · Forms Dataset: ChoiceGroupMIoU (%): 56.11 -> 83.01Improved
use ImageNet pretraining for DeepLabV3+ (ablation)DLV3+ · Forms Dataset: ChoiceGroup TitleMIoU (%): 18.69 -> 48.09Improved
use Lowresnet-TL instead of MSMT-FCN at stricter IoU 0.9MSMT-FCN · Marmot English (IoU 0.9)table detection F1 (%): 45.9 -> 62.8Improved
use Lowresnet-TL instead of MSMT-FCN at IoU 0.8MSMT-FCN · Marmot English (IoU 0.8)table detection F1 (%): 72.5 -> 73.7No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy