Dust, fallen leaves, or obstructions on solar panels are automatically identified as hazards by OpenMV

The photovoltaic panel “appears to be fine” does not necessarily indicate normal power generation efficiency. Localised shading and dust accumulation often persist for extended periods yet remain difficult to detect. OpenMV enables photovoltaic operation and maintenance to no longer rely entirely on manual inspections.

During the operation of photovoltaic power plants, foreign substances such as dust accumulation, fallen leaves, and bird droppings on the surface of photovoltaic panels directly affect light absorption efficiency, resulting in reduced power generation; over time, these issues may even accelerate component aging and shorten equipment service life.

For large-scale photovoltaic arrays, manual inspection is not only inefficient but also unable to promptly detect local anomalies. Leveraging SingTown’s OpenMV intelligent camera, long-term and stable visual monitoring of photovoltaic panel operational status can be achieved, enabling automatic identification and timely resolution of foreign object issues.

The camera can be fixedly deployed on a support structure to cover multiple photovoltaic module areas and continuously capture images of the photovoltaic panel surfaces. OpenMV performs region segmentation at the edge to delineate the effective operating areas of the photovoltaic panels and, by combining foreign object feature recognition with status comparison analysis, automatically determines whether abnormal dust accumulation, leaf obstruction, or localized contamination is present.

By comparing the current image with a normal reference, the system identifies abnormal areas affecting power generation efficiency and outputs the detection results. These results can directly trigger cleaning equipment, operation and maintenance systems, or alarm modules at the edge side, enabling immediate action upon detection without uploading large volumes of image data to the cloud.

Leveraging OpenMV’s edge computing capabilities and low-power design, this solution is particularly suitable for long-term deployment in large-scale, distributed photovoltaic scenarios.