Update README.md
Browse files
README.md
CHANGED
|
@@ -1,13 +1,81 @@
|
|
| 1 |
---
|
| 2 |
-
title: Art
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: 5.33.0
|
| 8 |
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
license: mit
|
|
|
|
|
|
|
|
|
|
| 11 |
---
|
| 12 |
|
| 13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
title: AI Art Historian
|
| 3 |
+
emoji: π¨
|
| 4 |
+
colorFrom: blue
|
| 5 |
+
colorTo: purple
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: 5.33.0
|
| 8 |
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
license: mit
|
| 11 |
+
short_description: Expert AI artwork analysis with historical insights
|
| 12 |
+
tags:
|
| 13 |
+
- agent-demo-track
|
| 14 |
---
|
| 15 |
|
| 16 |
+
# π¨ AI Art Historian
|
| 17 |
+
|
| 18 |
+
**Expert-level artwork analysis powered by Microsoft Phi-3.5-Vision and SmolAgent Framework**
|
| 19 |
+
|
| 20 |
+
## π How to use
|
| 21 |
+
|
| 22 |
+
1. **Upload an artwork** - Supports JPG, PNG and other image formats
|
| 23 |
+
2. **Ask questions** (optional) - Get specific analysis on style, symbols, or history
|
| 24 |
+
3. **Get expert analysis** - Receive detailed art historical insights in seconds
|
| 25 |
+
|
| 26 |
+
### Core Components
|
| 27 |
+
|
| 28 |
+
#### 1. **Vision Agent** - Phi-3.5-Vision-Instruct
|
| 29 |
+
- Converts artwork images to detailed descriptions
|
| 30 |
+
- Identifies visual elements, colors, composition, style indicators
|
| 31 |
+
|
| 32 |
+
#### 2. **Analysis Agent** - SmolAgent Framework
|
| 33 |
+
- Processes descriptions using specialized art tools
|
| 34 |
+
- Multi-step reasoning for comprehensive analysis
|
| 35 |
+
- Fallback system for reliability
|
| 36 |
+
|
| 37 |
+
## π§ Specialized Tools
|
| 38 |
+
|
| 39 |
+
### π¨ Style Detector
|
| 40 |
+
```python
|
| 41 |
+
@tool
|
| 42 |
+
def art_style_detector(visual_description: str) -> str:
|
| 43 |
+
"""Identifies artistic movements and styles"""
|
| 44 |
+
```
|
| 45 |
+
**Detects**: Renaissance, Baroque, Impressionism, Expressionism, Cubism, Surrealism, Abstract, Pop Art, Minimalism
|
| 46 |
+
|
| 47 |
+
### π Historical Context Provider
|
| 48 |
+
```python
|
| 49 |
+
@tool
|
| 50 |
+
def historical_context_provider(art_period: str) -> str:
|
| 51 |
+
"""Provides historical and cultural context"""
|
| 52 |
+
```
|
| 53 |
+
**Includes**: Time periods, key artists, cultural influences, movement characteristics
|
| 54 |
+
|
| 55 |
+
### π Symbolism Interpreter
|
| 56 |
+
```python
|
| 57 |
+
@tool
|
| 58 |
+
def symbolism_interpreter(visual_elements: str) -> str:
|
| 59 |
+
"""Decodes symbolic meanings in artwork"""
|
| 60 |
+
```
|
| 61 |
+
**Analyzes**: Colors (red, blue, gold), Objects (crown, flowers, skull), Religious symbols (cross, dove)
|
| 62 |
+
|
| 63 |
+
## β‘ Processing Workflow
|
| 64 |
+
|
| 65 |
+
1. **Image Analysis**: Phi-3.5-Vision describes artwork details
|
| 66 |
+
2. **Agent Reasoning**: SmolAgent processes description using tools:
|
| 67 |
+
- Style detection from visual cues
|
| 68 |
+
- Historical context for identified period
|
| 69 |
+
- Symbol interpretation from elements
|
| 70 |
+
3. **Structured Output**: Comprehensive art historical analysis
|
| 71 |
+
|
| 72 |
+
## π‘οΈ Robust Design
|
| 73 |
+
|
| 74 |
+
- **Dual Processing**: Vision model + reasoning agent
|
| 75 |
+
- **Fallback System**: Direct tool usage if agent fails
|
| 76 |
+
- **GPU Optimization**: Fast image processing
|
| 77 |
+
- **Dark Theme**: Eye-friendly interface
|
| 78 |
+
|
| 79 |
+
## π Acknowledgements
|
| 80 |
+
|
| 81 |
+
Made with β€οΈ by **Gizem Sarsinlar** ([@gizemsarsinlar](https://huggingface.co/gizemsarsinlar))
|