Solar energy is becoming a good alternative energy source as the world shifts towards renewable energy sources. To ensure perfect energy production and minimize different losses, it is essential to maintain efficient and regular monitoring and maintenance. The surfaces of the panels should be clean to receive a good amount of solar energy to be transformed. Defects on solar panels, such as dust accumulation, cracks, and bird droppings, can reduce the efficiency of the transformation, so there is a need for an automated and a real-time monitoring systems that can accurately detect these defects. This study represents a computer vision approach using YOLOv10 (You Only Look Once) deep learning variants to detect defects and find anomalies through cameras. Three different algorithm variants were used to locate different defect classes, with the Convolutional Neural Network (CNN) trained on 3,272 images to achieve detectable classes with a good detection percentage level.