Skip to content

Commit 7fffec1

Browse files
SarmaHighOnCodeSarmaHighOnCode
committed
fix: batch application of repository robustness and backend bugs
1 parent 0a9aaa8 commit 7fffec1

30 files changed

Lines changed: 9816 additions & 237 deletions

‎README.md‎

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -22,6 +22,15 @@
2222

2323
---
2424

25+
> [!IMPORTANT]
26+
> **Project Status Update — Core Infrastructure Complete, ML Model Pending**
27+
>
28+
> The vast majority of the backend, frontend, streaming architecture, and testing suite is fully implemented and operational. However, the core **PyTorch LSTM+MDN Handwriting Synthesis Model** has *not* been trained yet.
29+
> * The system currently defaults to `ENGINE_BACKEND=mock` in the `.env` file to simulate handwriting generation for testing.
30+
> * **Next step for full operation:** Run `scripts/download_iam.py` and `train.py` to generate the `.pt` checkpoint file, then switch the `.env` backend to `lstm`.
31+
32+
---
33+
2534
## Table of Contents
2635

2736
- [Executive Summary](#executive-summary)

‎SPACE_README.md‎

Lines changed: 54 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,54 @@
1+
---
2+
title: INKFORGE - Handwriting Synthesis
3+
emoji: ✍️
4+
colorFrom: blue
5+
colorTo: purple
6+
sdk: gradio
7+
sdk_version: 4.0.0
8+
app_file: app.py
9+
pinned: false
10+
license: mit
11+
tags:
12+
- handwriting
13+
- synthesis
14+
- lstm
15+
- generative
16+
- pytorch
17+
---
18+
19+
# INKFORGE - Handwriting Synthesis Engine
20+
21+
Generate realistic handwritten text using deep learning.
22+
23+
## Features
24+
25+
- **Multiple Writing Styles**: Neat cursive, casual print, rushed notes, and more
26+
- **Customizable Parameters**: Control variation, stroke width, ink color
27+
- **Full Document Generation**: Generate multi-paragraph handwritten documents
28+
- **Writing Fatigue Simulation**: Text degrades naturally over long passages
29+
30+
## How It Works
31+
32+
INKFORGE uses an LSTM + Mixture Density Network (MDN) architecture based on
33+
[Graves 2013](https://arxiv.org/abs/1308.0850) to generate handwriting stroke-by-stroke.
34+
35+
## Usage
36+
37+
1. Enter your text in the input box
38+
2. Select a writing style preset
39+
3. Adjust temperature (variation) and other parameters
40+
4. Click "Generate Handwriting"
41+
42+
## Note
43+
44+
This demo uses a randomly initialized model for demonstration.
45+
For realistic handwriting, train the model on the IAM On-Line Handwriting Database.
46+
47+
## Links
48+
49+
- [GitHub Repository](https://github.com/SarmaHighOnCode/INKFORGE)
50+
- [IAM Dataset](https://fki.tic.heia-fr.ch/databases/iam-on-line-handwriting-database)
51+
52+
## License
53+
54+
MIT License

0 commit comments

Comments
 (0)