River Monitoring System Based on Liquid Sensors and YOLOv8 on Raspberry Pi
The increase in human activities has caused various environmental problems, including river pollution that can lead to flooding and a decline in water quality. One of the main challenges is detecting and monitoring river conditions in real-time to ensure appropriate mitigation measures can be taken.
RiverSense is designed to provide an innovative solution by integrating liquid sensors, YOLOv8-based object recognition, and Raspberry Pi to monitor river water levels and detect waste on the water surface. This system aims to deliver up-to-date information to local users through a web-based server, assisting in quick and accurate decision-making.
RiverSense is a river condition monitoring system that integrates:
- Water level sensing using three Polypropylene Liquid Water Level Float Switches.
- Visual detection of waste in the river current using ESP32-CAM and the YOLOv8 Object Detection algorithm.
- All data is processed and controlled by the Raspberry Pi Model B+ and sent to a local HTML/JS-based website server.
The main objectives of RiverSense are to provide real-time information regarding:
- River water level
- The presence of waste detected on the river surface
The system utilizes a Mini UPS as a backup power supply to maintain operational stability during power outages.
- 📈 Real-time water level monitoring using three vertical float switches.
- 📷 Visual waste detection using an ESP32-CAM with the YOLOv8 model.
- 🌐 Automatic transmission of sensor data and detection results to a local website.
- ⚡ Backup power using a Mini UPS to keep the system running during power failures.
+-----------------+ +-----------------+ +-----------------+
| Liquid Float | | ESP32-CAM | | Raspberry Pi B+ |
| Sensors (3x) | | YOLO8 Detection | | Local Webserver |
+--------+--------+ +--------+--------+ +--------+--------+
| | |
+-----------+-------------+-----------+-------------+
|
Mini UPS
- Max Contact Rating: 10W
- Max Switching Voltage: 220V DC/AC
- Max Switching Current: 1.5A
- Temperature Rating: -10°C to +85°C
- Materials: Float Ball & Body - Polypropylene (PP)
- Dimensions: 23.3mm (Diameter) x 57.7mm (Height)
- Cable Length: 36cm
- Processor: Dual-core 32-bit, 240MHz
- Memory: 520KB SRAM + 4MB PSRAM
- Camera Support: OV2640 / OV7670
- WiFi Modes: STA / AP / STA+AP
- Storage: TF Card support
- Framework: FreeRTOS embedded
- Processor: ARM1176JZF-S, 700MHz
- RAM: 512MB SDRAM
- Connectivity: WiFi Dongle USB
- Storage: microSD
Liquid Float Sensors -> GPIO Raspberry Pi
ESP32-CAM -> WiFi Stream (Captured Image) -> Raspberry Pi
Raspberry Pi -> YOLO8 Inference
Raspberry Pi -> Data Aggregation -> Local Website (HTML/JS)
- Data is sent in JSON format to the local server via HTTP POST.
- The local website polls or receives push updates to refresh data dynamically.
- Mount the 3x Liquid Water Level Sensors vertically in a waterproof protective box.
- Connect the three sensors to the GPIO of the Raspberry Pi Model B+.
- Prepare the ESP32-CAM to monitor the river, ensuring stable WiFi coverage to the Raspberry Pi.
- Connect the Raspberry Pi and ESP32-CAM to the Mini UPS.
- Flash the ESP32-CAM with firmware for image streaming via HTTP.
- Deploy the YOLOv8 model to the Raspberry Pi using Python (recommendation: Ultralytics YOLOv8).
- Run the detection and sensor monitoring scripts on the Raspberry Pi.
- Run the local HTML/JS-based server for data visualization.
- Open a browser and access the local website (example:
http://192.168.1.100/). - Monitor the river water level and waste detection in real-time.
- The latest data will be updated automatically without manual refresh.
- Telegram-based notification integration when the water level exceeds safe limits.
- Historical data storage to a local database (SQLite).
- Visualization of water level graphs and the number of detected waste items.