sci-figure

Name: sci-figure
Author: ShZhao27208

175installs23forks

How to install sci-figure?

shzhao27208/aut_sci_write/sci-figure

$npx -y skills add ShZhao27208/Aut_Sci_Write --skill sci-figure

Installs into the current project.

Summary

Extracts figures and sub-figures from academic PDF papers. Supports Fig/Figure, Scheme, Chart, Supplementary Figure, Extended Data Figure (Nature), and Chinese equivalents (图/方案/示意图/附图/补充图). Sub-figure label recognition supports (a)/(A)/a)/(i)/(1)/a. formats. High-quality PNG out

Files · 1

View on GitHub

SKILL.md

1	# Sci-Figure — Scientific Figure Extractor
2
3	Precisely extract figures and sub-figures from academic PDF papers.
4
5	> License note: sci-figure is licensed under AGPL-3.0-or-later because it links [PyMuPDF (fitz)](https://pymupdf.readthedocs.io/), which is AGPL-licensed.
6
7	## Installation
8
9	Install the package from the skill directory before first use:
10
11	```bash
12	cd ${SKILL_DIR}
13	pip install -e .
14	```
15
16	This registers the `sh-sci-fig` CLI command. Requires Tesseract OCR:
17	- Windows: `winget install UB-Mannheim.TesseractOCR`
18	- Linux: `apt install tesseract-ocr`
19	- macOS: `brew install tesseract`
20
21	## Preferences (EXTEND.md)
22
23	Use Bash to check EXTEND.md existence (priority order):
24
25	```bash
26	# Check project-level first
27	test -f .baoyu-skills/sci-figure/EXTEND.md && echo "project"
28
29	# Then user-level (cross-platform: $HOME works on macOS/Linux/WSL)
30	test -f "$HOME/.baoyu-skills/sci-figure/EXTEND.md" && echo "user"
31	```
32
33	EXTEND.md Supports: Default DPI \| Default output format \| Tesseract path
34
35	## Usage
36
37	```bash
38	sh-sci-fig <input.pdf> [options]
39	```
40
41	## Options
42
43	\| Option \| Short \| Description \| Default \|
44	\|--------\|-------\|-------------\|---------\|
45	\| `<input>` \| \| PDF file path \| Required \|
46	\| `--figure` \| `-f` \| Figure number (1, 2, 3...) \| Required (except --list/--all) \|
47	\| `--subfigure` \| `-s` \| Sub-figure label (a, b, c...) \| None (returns whole figure) \|
48	\| `--output` \| `-o` \| Output directory \| Current directory \|
49	\| `--dpi` \| `-d` \| Output resolution \| 600 \|
50	\| `--list` \| `-l` \| List all available figure numbers \| false \|
51	\| `--all` \| \| Extract all figures \| false \|
52	\| `--format` \| \| Output format (png/jpg) \| png \|
53	\| `--strategy` \| \| Extraction strategy: hybrid/native/cv \| hybrid \|
54	\| `--ocr` \| \| OCR engine: tesseract/easyocr/none \| tesseract \|
55	\| `--render-page` \| \| Render full page with annotations \| false \|
56	\| `--annotate` \| \| Draw bounding boxes on rendered page \| false \|
57	\| `--bbox` \| \| Manual bbox override (x0,y0,x1,y1 in px) \| None \|
58	\| `--no-trim` \| \| Disable whitespace trimming \| false \|
59	\| `--debug` \| \| Enable debug logging \| false \|
60	\| `--quiet` \| `-q` \| Suppress info messages \| false \|
61
62	## Examples
63
64	```bash
65	# Extract Figure 2, sub-figure c
66	sh-sci-fig paper.pdf -f 2 -s c
67
68	# Extract entire Figure 3
69	sh-sci-fig paper.pdf -f 3
70
71	# List all available figures in a PDF
72	sh-sci-fig paper.pdf --list
73
74	# Extract all figures
75	sh-sci-fig paper.pdf --all
76
77	# Custom output directory and DPI
78	sh-sci-fig paper.pdf -f 2 -s c -o ./output/ -d 300
79
80	# Use EasyOCR for sub-figure label detection
81	sh-sci-fig paper.pdf --all --ocr easyocr
82
83	# CV-only strategy (skip native extraction)
84	sh-sci-fig paper.pdf --all --strategy cv
85
86	# Render page with annotated bounding boxes (debugging)
87	sh-sci-fig paper.pdf -f 1 --render-page --annotate
88
89	# Manual bbox extraction (multimodal correction)
90	sh-sci-fig paper.pdf -f 1 --bbox 100,200,800,1200
91	```
92
93	Output:
94	```
95	Extracted: figure_2c.png (1920x1080, 600 DPI)
96	```
97
98	## Error Handling
99
100	\| Scenario \| Behavior \|
101	\|----------\|----------\|
102	\| Figure number not found \| Error + list all available figure numbers \|
103	\| OCR recognition failed \| Return entire figure region \|
104	\| Sub-figure split failed \| Return entire figure region \|
105	\| No sub-figure labels found \| Return entire figure region \|
106
107	## Tech Stack
108
109	\| Library \| Role \|
110	\|---------\|------\|
111	\| pdfplumber \| Text + coordinate extraction (caption detection) \|
112	\| PyMuPDF (fitz) \| Native image extraction + high-quality page rendering \|
113	\| opencv-python \| CV region detection, connected-component analysis, content validation \|
114	\| Pillow \| Final cropping, format conversion \|
115	\| pytesseract \| OCR for sub-figure label recognition (default) \|
116	\| easyocr \| Alternative OCR engine (optional, `pip install sci-figure[ocr]`) \|
117	\| numpy \| Image array operations \|
118
119	## Extraction Engines (v2)
120
121	\| Engine \| Priority \| Best For \|
122	\|--------\|----------\|----------\|
123	\| Native (PyMuPDF) \| 1st \| Raster images embedded in PDF \|
124	\| CV (connected-component) \| 2nd \| Vector graphics, colored plots \|
125	\| Caption-anchored \| 3rd \| Fallback when above engines fail \|
126
127	The `hybrid` strategy (default) tries all three in order and validates results.
128
129	## Detected Figure Fields
130
131	Each figure returned by `FigureExtractor.detect_al

People who liked this also viewed

6 picks

Type

sci-figure

175installs23forks

View on GitHub

How to install sci-figure?

shzhao27208/aut_sci_write/sci-figure

$npx -y skills add ShZhao27208/Aut_Sci_Write --skill sci-figure

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Sci-Figure — Scientific Figure Extractor
2
3	Precisely extract figures and sub-figures from academic PDF papers.
4
5	> License note: sci-figure is licensed under AGPL-3.0-or-later because it links [PyMuPDF (fitz)](https://pymupdf.readthedocs.io/), which is AGPL-licensed.
6
7	## Installation
8
9	Install the package from the skill directory before first use:
10
11	```bash
12	cd ${SKILL_DIR}
13	pip install -e .
14	```
15
16	This registers the `sh-sci-fig` CLI command. Requires Tesseract OCR:
17	- Windows: `winget install UB-Mannheim.TesseractOCR`
18	- Linux: `apt install tesseract-ocr`
19	- macOS: `brew install tesseract`
20
21	## Preferences (EXTEND.md)
22
23	Use Bash to check EXTEND.md existence (priority order):
24
25	```bash
26	# Check project-level first
27	test -f .baoyu-skills/sci-figure/EXTEND.md && echo "project"
28
29	# Then user-level (cross-platform: $HOME works on macOS/Linux/WSL)
30	test -f "$HOME/.baoyu-skills/sci-figure/EXTEND.md" && echo "user"
31	```
32
33	EXTEND.md Supports: Default DPI \| Default output format \| Tesseract path
34
35	## Usage
36
37	```bash
38	sh-sci-fig <input.pdf> [options]
39	```
40
41	## Options
42
43	\| Option \| Short \| Description \| Default \|
44	\|--------\|-------\|-------------\|---------\|
45	\| `<input>` \| \| PDF file path \| Required \|
46	\| `--figure` \| `-f` \| Figure number (1, 2, 3...) \| Required (except --list/--all) \|
47	\| `--subfigure` \| `-s` \| Sub-figure label (a, b, c...) \| None (returns whole figure) \|
48	\| `--output` \| `-o` \| Output directory \| Current directory \|
49	\| `--dpi` \| `-d` \| Output resolution \| 600 \|
50	\| `--list` \| `-l` \| List all available figure numbers \| false \|
51	\| `--all` \| \| Extract all figures \| false \|
52	\| `--format` \| \| Output format (png/jpg) \| png \|
53	\| `--strategy` \| \| Extraction strategy: hybrid/native/cv \| hybrid \|
54	\| `--ocr` \| \| OCR engine: tesseract/easyocr/none \| tesseract \|
55	\| `--render-page` \| \| Render full page with annotations \| false \|
56	\| `--annotate` \| \| Draw bounding boxes on rendered page \| false \|
57	\| `--bbox` \| \| Manual bbox override (x0,y0,x1,y1 in px) \| None \|
58	\| `--no-trim` \| \| Disable whitespace trimming \| false \|
59	\| `--debug` \| \| Enable debug logging \| false \|
60	\| `--quiet` \| `-q` \| Suppress info messages \| false \|
61
62	## Examples
63
64	```bash
65	# Extract Figure 2, sub-figure c
66	sh-sci-fig paper.pdf -f 2 -s c
67
68	# Extract entire Figure 3
69	sh-sci-fig paper.pdf -f 3
70
71	# List all available figures in a PDF
72	sh-sci-fig paper.pdf --list
73
74	# Extract all figures
75	sh-sci-fig paper.pdf --all
76
77	# Custom output directory and DPI
78	sh-sci-fig paper.pdf -f 2 -s c -o ./output/ -d 300
79
80	# Use EasyOCR for sub-figure label detection
81	sh-sci-fig paper.pdf --all --ocr easyocr
82
83	# CV-only strategy (skip native extraction)
84	sh-sci-fig paper.pdf --all --strategy cv
85
86	# Render page with annotated bounding boxes (debugging)
87	sh-sci-fig paper.pdf -f 1 --render-page --annotate
88
89	# Manual bbox extraction (multimodal correction)
90	sh-sci-fig paper.pdf -f 1 --bbox 100,200,800,1200
91	```
92
93	Output:
94	```
95	Extracted: figure_2c.png (1920x1080, 600 DPI)
96	```
97
98	## Error Handling
99
100	\| Scenario \| Behavior \|
101	\|----------\|----------\|
102	\| Figure number not found \| Error + list all available figure numbers \|
103	\| OCR recognition failed \| Return entire figure region \|
104	\| Sub-figure split failed \| Return entire figure region \|
105	\| No sub-figure labels found \| Return entire figure region \|
106
107	## Tech Stack
108
109	\| Library \| Role \|
110	\|---------\|------\|
111	\| pdfplumber \| Text + coordinate extraction (caption detection) \|
112	\| PyMuPDF (fitz) \| Native image extraction + high-quality page rendering \|
113	\| opencv-python \| CV region detection, connected-component analysis, content validation \|
114	\| Pillow \| Final cropping, format conversion \|
115	\| pytesseract \| OCR for sub-figure label recognition (default) \|
116	\| easyocr \| Alternative OCR engine (optional, `pip install sci-figure[ocr]`) \|
117	\| numpy \| Image array operations \|
118
119	## Extraction Engines (v2)
120
121	\| Engine \| Priority \| Best For \|
122	\|--------\|----------\|----------\|
123	\| Native (PyMuPDF) \| 1st \| Raster images embedded in PDF \|
124	\| CV (connected-component) \| 2nd \| Vector graphics, colored plots \|
125	\| Caption-anchored \| 3rd \| Fallback when above engines fail \|
126
127	The `hybrid` strategy (default) tries all three in order and validates results.
128
129	## Detected Figure Fields
130
131	Each figure returned by `FigureExtractor.detect_al

People who liked this also viewed

6 picks

Type