![]() ![]() ![]() ![]() The model is trained on TF1 SSD Mobilenet V1 using WSL along with MobaXterm (WSL and MobaXterm are not actually required as it was just matter of personal preference). The model has been trained by transfer learning techniques using CPU trained coco 300x300 model which was trained on large dataset of different animal images, this particular model used for disease detection also used hundreds of annotated images (with corresponding bounding boxes) of different diseases and also healthy leaves. The tomato leaf disease object detection model is the pre-trained deep learning model, specifically designed for the detection of common tomato leaf diseases namely Bacterial Spot, Early Blight, Late Blight, Septoria Spot and Yellow Leaf Curl diseases. As Google Coral Dev Board is resource scarce (in terms of using relatively low power) emmbedded device it has relatively small yet powerful processing capability using TPU (Tensor Processing Unit) with 8mb sram allowing us to perform real-time inference on tomato leaf images, possibly enabling early detection of diseases and potentially helping farmers and improving crop yield. This project aims to detect 5 different tomato leaf diseases and 6 different class inlcuding Healthy leaf using a Coral Edge TPU Dev Board. Tomato leaf disease inference using the Coral Edge TPU Dev Board Tomato Leaf Disease Object Detection Model using Coral Edge TPU Dev Board Introduction ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |