Spaces:
Sleeping
Sleeping
initial files
Browse files- src/README.md +32 -0
- src/requirements.txt +6 -0
src/README.md
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
title: ShopGuard AI
|
| 3 |
+
emoji: 🛡️
|
| 4 |
+
colorFrom: blue
|
| 5 |
+
colorTo: red
|
| 6 |
+
sdk: streamlit
|
| 7 |
+
sdk_version: 1.32.0
|
| 8 |
+
app_file: app.py
|
| 9 |
+
pinned: false
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# 🛡️ ShopGuard AI — Shoplifting Detection System
|
| 13 |
+
|
| 14 |
+
FYP Demo | YOLO11n → MobileNetV2 → Attention LSTM pipeline
|
| 15 |
+
|
| 16 |
+
## Models
|
| 17 |
+
| Model | HuggingFace Repo | Scene |
|
| 18 |
+
|-------|-----------------|-------|
|
| 19 |
+
| Model A | higsboson/shoplifting_exp_a | General |
|
| 20 |
+
| Model B | higsboson/shoplifting_exp_b | Kitchen |
|
| 21 |
+
| Model C | higsboson/shoplifting_exp_c | Lab |
|
| 22 |
+
|
| 23 |
+
## How to Use
|
| 24 |
+
1. Select a model from the dropdown
|
| 25 |
+
2. Adjust the decision threshold slider
|
| 26 |
+
3. Upload a short video clip (mp4/avi/mov/mkv)
|
| 27 |
+
4. Click **Run Inference**
|
| 28 |
+
|
| 29 |
+
## Pipeline
|
| 30 |
+
- **YOLO11n** detects and crops the person in each frame
|
| 31 |
+
- **MobileNetV2** extracts 1280-dim features per frame
|
| 32 |
+
- **Attention LSTM** classifies the sequence as Shoplifting or Normal
|
src/requirements.txt
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
streamlit>=1.32.0
|
| 2 |
+
tensorflow>=2.15.0
|
| 3 |
+
numpy>=1.24.0
|
| 4 |
+
opencv-python-headless>=4.8.0
|
| 5 |
+
ultralytics>=8.0.0
|
| 6 |
+
huggingface_hub>=0.21.0
|