Real-Time Seed Detection and Germination Analysis in Precision Agriculture: A Fusion Model With U-Net and CNN on Jetson Nano
D. Ramesh Reddy, Ramalingaswamy Cheruku, Prakash Kodali
Abstract
Precision agriculture involves the strategic utilization of resources, precise application of inputs, and continuous monitoring of crop health with the aim of enhancing productivity and sustainability in the field of agriculture. However, seed quality is difficult since natural differences among seed batches may affect germination rates, vigor, and crop performance. Hence, in this article, a novel fusion model for seed detection and germination is proposed. The proposed model combines the U-Net and CNN architectures for seed segmentation and classification, respectively. By harnessing U-Net's capabilities in image segmentation and CNN's strengths in classification, the proposed approach enables effective seed germination analysis. In addition, the model is specifically optimized for real-time processing and applications by implementing it on the NVIDIA Jetson Nano embedded GPU platform. The proposed fusion model achieved 0.91 pixel accuracy, 0.84 intersection over union, and 0.90 precision. The proposed model demonstrated excellent predictive ability when compared with the ResNet50, Inception, and LeNet. In addition, the proposed model requires less number of trainable parameters after LeNet. Further, the proposed model tested in real time and achieved 0.26 ms latency.