Evaluation of seed-metering devices using an automated system based on digital image analysis

The quality of planting directly influences crop productivity and is evaluated based on planting density and the uniform distribution of seeds along the furrows. To ensure consistent seed spacing, planters are equipped with seed-metering devices (seed meters ) that release seeds in a controlled manner. To assess the performance of such planters, digital image analysis methods have been developed, utilizing high-speed cameras and high-performance processing systems. This study presents an automated system to assess the performance of seed meters. The system includes a camera capable of up to 90 FPS, a designed diffuse dome light and a developed algorithm executed on a Raspberry Pi 3. The system was evaluated in three steps: (1) comparison between the distances detected by the system and reference distances; (2) a 3 2 factorial design considering conveyor belt speed (3, 5, and 7 km/h) and the camera’s frame rate (30, 60, and 90 FPS) as factors; and (3) assessment of performance parameters following ISO 7256/1 standard. Once validated, the system was applied to a case study. The system achieved an accuracy of 98.65% using video recording at 60 frames per second and a conveyor speed of 5 km/h, and demonstrated the ability to calculate the ISO-7256/1 parameters of the seed meter reaching 99.77% accuracy. The proposed system represents a versatile tool for the development of new seed meters and the optimization of existing devices by enabling accurate and automated performance evaluation.