Add font style information to the `Word` class
Maintainers usually reply within 1 day
Nobody has claimed this yet.
Assessment
- Difficulty
- 5/5
- Estimated time
- Over a week
- Newbie friendliness
- 12/100
- Issue type
- Feature
- Clarity
- Needs clarification
- Activity status
- Stale
- Tech stack
- python, pytorch, tensorflow
- Domain
- machine-learning
Research direction
The request is to add font style detection (bold, colour, underline) to the Word class, but the issue names no file or entry point. Look at where Word is defined in the OCR predictor output and at how the recognition models produce per-word results. Done would mean agreeing on the style schema and a model or post-processing step, which is design work for maintainers first.
Written by the indexing model from the issue text.
Description
🚀 The feature
I use docTR as OCR pre-processing before I send the text data into a LLM to extract data. However, a lot of information is encoded in font style like important things are often in bold or red or double underlined. Since there is already the possibility to combine networks, I was wondering whether you can train/add a network which can estimate font styles.
Motivation, pitch
Using annotated data for LLMs increases the accuracy of the task the LLM has to do, because if gets a context what is the important thing in a line or block.
Alternatives
Currently there is no alternative then just hoping the LLM can figure it out.
Additional context
No response
- Dominant language
- Python
- Stars
- 6.4k
- Forks
- 684
- Avg merge
- 20h 40m
- Merged PRs (30d)
- 30
Getting set up
- Ships a Dockerfile or Docker Compose file
- No pull request template
- Read the contributing guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from mindee/doctr
-
ext: tests topic: ci type: enhancement
Difficulty 3/5 Half a day Newbie friendliness 45/100
mindee/doctr#2171 · 1 comment ·
Maintainers usually reply within 1 day
-
Per-task YAML configuration for the reference training scriptsPossibly taken @felixdittrich92 claimed this 22 days ago. Openext: references topic: documentation type: new feature
mindee/doctr#2149 · 1 comment · 1 assignee ·
Maintainers usually reply within 1 day
-
Release tracker - v1.2.0May be free again @felixdittrich92 claimed this 64 days ago, and no pull request is open. Opencritical
mindee/doctr#2126 · 1 assignee ·
Maintainers usually reply within 1 day
-
[table structure] Extend table structure datasetMay be free again @simont2k claimed this 85 days ago, and no pull request is open. Opentopic: table comprehension
mindee/doctr#2119 · 1 comment · 1 assignee ·
Maintainers usually reply within 1 day
-
Difficulty 4/5 3-5 days Newbie friendliness 15/100
Maintainers usually reply within 1 day
Similar issues
-
Uninstall guide omits the CloudWatch log groups AWS CodeBuild creates for useCodeBuild pipelinesOpenbug documentation
Difficulty 1/5 1-3 hours Newbie friendliness 88/100
awslabs/visual-asset-management-system#552 ·
Maintainers usually reply within 1 day
-
netbox status: needs triage type: documentation
Difficulty 1/5 Under an hour Newbie friendliness 76/100
netbox-community/netbox#23460 ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 75/100
awslabs/threat-composer#362 ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
Maintainers usually reply within 1 day
-
bug
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
ansys/pyansys-common-mcp#213 ·
Maintainers usually reply within 3 days