Efficient management of irrigation water in agricultural communities requires reliable data on the level and flow rate in the distribution canals. Traditional methods, based on manual measurements or intrusive sensors, have limitations due to their cost, complexity of implementation, and lack of continuity in records. This paper proposes an IoT system with computer vision for non-intrusive monitoring of community irrigation canals. The solution uses a Raspberry Pi 5 together with a USB camera with night vision and Heltec WiFi LoRa 32 (V2) modules, which transmit level and flow data every 10 min using the MQTT protocol to a Node-RED environment. From there, the information is recorded in an online spreadsheet and integrated with Power BI, both for real-time visualization using the streaming dataset functionality and for the creation of interactive reports based on the data obtained. In addition, a bot was implemented on Telegram to send periodic notifications and a photovoltaic system was installed to ensure energy autonomy. The experimental results show that the system allows irrigation demand patterns to be identified according to times and days, providing objective information for community decision making and promoting more efficient and sustainable water resource management.