Efficient-AgriNet: A Hybrid Deep Learning Model for Crops Disease Detection

Bangladesh is primarily an agricultural country, where major crops such as rice, corn, potatoes, and wheat form the backbone of the economy. However, these crops are increasingly vulnerable to various threats, including viruses, bacterial infections, climate change, and other natural or artificial factors, leading to complex and widespread crop diseases. Motivated by the urgent need to support farmers with reliable and accessible technological solutions, this research sets out to develop an efficient and accurate disease detection system specifically tailored to the agricultural context of Bangladesh. The primary objectives are identifying crop diseases early, reducing crop losses, and improving sustainable agriculture through artificial intelligence. This work proposes a novel hybrid deep learning model named Efficient-AgriNet, which combines a pretrained EfficientNetB0 backbone with additional custom layers designed to optimize disease classification performance. The model was trained and evaluated using the Bangladeshi Crops Disease Dataset. Experimental results demonstrate that Efficient-AgriNet achieved an outstanding 97% accuracy and F1-score, confirming its ability to classify a wide range of crop diseases accurately. In conclusion, the proposed model provides a practical, high-performing solution that can empower farmers and agricultural practitioners by enabling early diagnosis, informed decision-making, and improved crop productivity in the face of evolving agricultural challenges. Further research and development of the Efficient-AgriNet model could lead to even more accurate and efficient disease detection systems, opening up new possibilities for the future of agriculture in Bangladesh.