vesuvius-patch-agg-analysis

Vesuvius challenge analysis of patch aggregation methods in ink detection

https://github.com/lschlessinger1/vesuvius-patch-agg-analysis

Science Score: 67.0%

This score indicates how likely this project is to be science-related based on various indicators:

✓
CITATION.cff file
Found CITATION.cff file
✓
codemeta.json file
Found codemeta.json file
✓
.zenodo.json file
Found .zenodo.json file
✓
DOI references
Found 2 DOI reference(s) in README
✓
Academic publication links
Links to: zenodo.org
○
Academic email domains
○
Institutional organization owner
○
JOSS paper metadata
○
Scientific vocabulary similarity
Low similarity (7.0%) to scientific vocabulary

Last synced: 6 months ago · JSON representation ·

Repository

Vesuvius challenge analysis of patch aggregation methods in ink detection

Basic Info

Host: GitHub
Owner: lschlessinger1
License: mit
Default Branch: main
Size: 3.1 MB

Statistics

Stars: 0
Watchers: 1
Forks: 0
Open Issues: 0
Releases: 1

Created about 1 year ago · Last pushed about 1 year ago

Metadata Files

Readme License Citation

Analysis of Patch Aggregation Methods in Ink Detection

Vesuvius Challenge analysis of patch aggregation methods in ink detection

Overview

This repository contains an analysis of different patch aggregation methods used for ink detection in the Vesuvius Challenge. The analysis explores various approaches to merging patch-based predictions into a reconstructed ink prediction.

The full analysis is available in the PDF: Analysisofpatchaggregationmethodsinink_detection.pdf

Open in Google Colab

To interact with the analysis notebooks, click the corresponding button below:

Generate and Save Patch Aggregations:
Quantitative Analysis:
Qualitative Analysis:

Summary of Methods

The following patch aggregation techniques were evaluated: - Averaging – Each patch contributes equally to the final reconstructed image, reducing bias towards any particular pixel. - Gaussian – A 2D Gaussian window emphasizes the central part of each patch, reducing boundary artifacts and smoothing transitions. - Cropping – Center-cropping overlapping patch predictions before averaging, reducing the impact of window edges and corners. - Hanning – A 2D Hann window emphasizes the center of each patch while smoothly tapering at the edges.

Results

Evaluation Metrics: The effectiveness of different methods was quantified using Average Precision (AP) and F0.5 score.
Visualization: The repository includes qualitative comparisons of aggregated results vs. ground truth, analyzing the impact of stride, window size, and patch aggregation method on the final predictions.

Owner

Name: Lou Schlessinger
Login: lschlessinger1
Kind: user
Location: Philadelphia

Website: http://louschlessinger.com/
Twitter: louschlessinger
Repositories: 6
Profile: https://github.com/lschlessinger1

Citation (CITATION.cff)

cff-version: 1.2.0
title: vesuvius-patch-agg-analysis
message: >-
  If you use this software, please cite it using the
  metadata from this file.
type: software
authors:
  - given-names: Louis
    family-names: Schlessinger
  - given-names: Arefeh
    family-names: Sherafati
url: >-
  https://github.com/lschlessinger1/vesuvius-patch-agg-analysis
abstract: >-
  Vesuvius Challenge analysis of patch aggregation methods
  in ink detection
keywords:
  - vesuvius-challenge
  - ink-detection
license: MIT

ecosyste.ms

Data

Tools

Indexes

Applications

Experiments

Open Source Science