Spaces:
Sleeping
Sleeping
Btr4k
commited on
Commit
·
1780f24
1
Parent(s):
f77f85c
Fix README configuration
Browse files
README.md
DELETED
@@ -1,37 +0,0 @@
|
|
1 |
-
# 📧 Phishing Detector (Text + Screenshot)
|
2 |
-
|
3 |
-
This Hugging Face Space provides an interface to analyze emails and SMS messages for potential phishing attempts. It accepts both direct text input and screenshots, using OCR to extract text from images when needed.
|
4 |
-
|
5 |
-
## Features
|
6 |
-
|
7 |
-
- Support for both text input and screenshot upload
|
8 |
-
- OCR for processing screenshots (microsoft/trocr-base-printed)
|
9 |
-
- Zero-shot classification to detect phishing patterns
|
10 |
-
- Keyword analysis for common phishing indicators
|
11 |
-
- Confidence scores for classifications
|
12 |
-
- Detailed explanations of suspicious elements
|
13 |
-
- Example messages for demonstration
|
14 |
-
|
15 |
-
## How to Use
|
16 |
-
|
17 |
-
1. Either:
|
18 |
-
- Paste your message text into the input box, OR
|
19 |
-
- Upload a screenshot of the message
|
20 |
-
2. Click "Submit" to analyze
|
21 |
-
3. Review the classification and explanation
|
22 |
-
|
23 |
-
## Technical Details
|
24 |
-
|
25 |
-
- Uses microsoft/trocr-base-printed for OCR
|
26 |
-
- Uses facebook/bart-large-mnli for zero-shot classification
|
27 |
-
- Implements keyword-based heuristics
|
28 |
-
- Built with Gradio for the user interface
|
29 |
-
- No fine-tuning required
|
30 |
-
|
31 |
-
## Disclaimer
|
32 |
-
|
33 |
-
This is an educational demo and should not be used as your only method of detecting phishing attempts. Always exercise caution with suspicious messages and follow your organization's security guidelines.
|
34 |
-
|
35 |
-
## License
|
36 |
-
|
37 |
-
MIT License
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|