A smart trash-sorting prototype built for the Samsung Innovation Campus Hackathon, capable of sorting waste into categories in real time using a custom-trained CNN model and ESP32 microcontroller.
- See our streamlit web here: https://shijinforgesmartbin.streamlit.app/
- See our Youtube Final Pitching & Demo here: https://youtu.be/JbN-0TZBvRU
- See our Youtube Code Explanation here: https://youtu.be/25jmLoOJSeo
- AI-based waste classification (Organic, Plastic, Others)
- Custom-trained CNN model with 45,000 image samples
- Servo control and ESP32 camera integration
- Real-time motion detection & sorting in <2 seconds
- Streamlit interface for demonstration
- Backend: Python, Flask, Streamlit
- AI Model: TensorFlow, Keras, MobileNetV2
- IoT: ESP32, MicroPython, Servo, Motion Sensors
- Others: GitHub for collaboration, iCrawler for scraping dataset
- My Role: Backend engineer, API architect, and streamlit visualization
- Helped lead the system integration and workflow planning
- My first time working with deep learning and ESP32 microcontrollers
- Learned to optimize a smart system end-to-end: AI → Hardware → UX
- Overcame challenges in real-time performance and hardware-software communication
- Gained leadership and collaboration skills using GitHub in a real team setting
Quarter-finalist at Samsung Innovation Campus Batch 6 Hackathon
Apache 2.0 – educational purposes only