How ahmednasr6770164-ui/crash-detection-cnn works
This repository presents an accident/no-accident image classifier, with a Streamlit upload-and-predict workflow and a bundled EfficientNetB0 model. The README also advertises video-based real-time detection, Grad-CAM checks, and email alerts; these are shown as documented capabilities, but their implementation and wiring cannot be confirmed from the sampled source. The notebooks indicate model-development work, but their contents were not inspected.
Main entry points: app.py Streamlit app · best_crash_detection_efficientnetb0.keras EfficientNetB0 model · README.md Real-time video detection
Components 15
Image inference
- Streamlit app
streamlit/app.py - Image upload
streamlit/app.py - Image preprocessing
streamlit/app.py - Image prediction
streamlit/app.py - Prediction results
streamlit/app.py
Model assets
- Model loader
streamlit/app.py - EfficientNetB0 model
streamlit/best_crash_detection_efficientnetb0.keras - CNN notebook
Accident_Detection_CNN_(2).ipynb - EfficientNet notebook
efficientnet_headRegulaization_headsize.ipynb - Crash detection notebook
ahmed_crash_detection_cnn.ipynb
Documented capabilities
- Real-time video detection
README.md - Grad-CAM checks
README.md - Automated email alerts
README.md
Other components
- User
- Email service
Connections 13
- User → Streamlit app: uses
- Streamlit app → Model loader: loads model
- Model loader → EfficientNetB0 model: reads
- User → Image upload: uploads image
- Image upload → Streamlit app: provides image
- Streamlit app → Image prediction: invokes
- Image prediction → Image preprocessing: preprocesses
- Image prediction → EfficientNetB0 model: runs inference
- Image prediction → Prediction results: returns prediction
- Streamlit app → Prediction results: displays
- Real-time video detection → EfficientNetB0 model: uses model
- Grad-CAM checks → EfficientNetB0 model: checks model
- Automated email alerts → Email service: sends alerts