File size: 2,717 Bytes
561e703
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2003d2f
 
 
561e703
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
---
title: Road Sign Recognition System
emoji: 🚦
colorFrom: red
colorTo: yellow
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
license: mit
---

# 🚦 Road Sign Recognition System

## Overview
An intelligent computer vision application that identifies and classifies road signs from images using AutoGluon's MultiModal Predictor. This system helps with driver education, navigation assistance, and road safety analysis.

## Model Details
- **Model Type**: AutoGluon MultiModal Predictor
- **Architecture**: Deep learning with vision transformer backbone
- **Source Model**: [its-zion-18/sign-image-autogluon-predictor](https://huggingface.co/its-zion-18/sign-image-autogluon-predictor)
- **Task**: Multi-class image classification for road sign detection
- **Training**: AutoGluon automatic model selection and hyperparameter tuning
- 
## Limitations
- **Not Enough Classes** Currently only have 2 classes (Stop and Speed limit)

## Features

### Input Methods
- **File Upload**: Support for JPG, PNG, and other common image formats
- **Webcam Capture**: Real-time photo capture for immediate classification
- **Clipboard Paste**: Quick image input from clipboard
- **Example Images**: Pre-loaded road sign samples for testing

### Output Information
- **Primary Classification**: Most likely sign type with confidence score
- **Top Predictions**: Alternative classifications with probabilities
- **Confidence Visualization**: Progress bars showing probability distribution
- **Detailed Analysis**: Interpretation of confidence levels

## How to Use

1. **Upload or Capture**: Provide an image of a road sign
2. **Automatic Processing**: Model analyzes the sign immediately
3. **View Results**: See the classification and confidence scores
4. **Try Examples**: Test with provided sample images

## Best Practices
- Use clear, well-focused images
- Ensure the entire sign is visible
- Avoid excessive angles or distortion
- Good lighting improves accuracy
- One sign per image works best

## Technical Specifications
- **Framework**: AutoGluon for automated machine learning
- **Backend**: PyTorch for deep learning operations
- **Deployment**: Hugging Face Spaces
- **Interface**: Gradio for user interaction

## Applications
- Driver education and training
- Navigation system enhancement
- Municipal sign inventory
- Road safety assessment
- Autonomous vehicle development

## Limitations
- Optimized for standard road signs
- Single sign detection per image
- May vary with weather conditions
- Region-specific signs may not be recognized

## Privacy
- No image storage after processing
- No personal data collection
- Server-side processing only

---
*Created for CMU 24-679 Course Assignment*