Crop Disease Diagnosis with Deep Learning-Based Image Captioning and Object Detection

Urban Farm DB summary

This study proposes a deep learning-based crop disease diagnosis solution for novice urban farmers, integrating Image Captioning and Object Detection models. The Image Captioning model, using InceptionV3 and Transformer, achieved an average BLEU score of 64.96%, demonstrating high performance in generating detailed diagnostic sentences. The Object Detection model, employing YOLOv5, achieved an mAP50 of 0.382, indicating a need for further improvement in detecting infected areas.