Identification of Weeds in Wheat Crop Using Artificial Intelligence Techniques
Harsh Sachan *
ICAR-Indian Agricultural Statistics Research Institute, New Delhi, 110 012, India.
S. N. Islam
ICAR-Indian Agricultural Statistics Research Institute, New Delhi, 110 012, India.
Shivadhar Misra
ICAR- Indian Agricultural Research Institute, New Delhi, India.
Sudeep Marwaha
ICAR-Indian Agricultural Statistics Research Institute, New Delhi, 110 012, India.
Ashraful Haque
ICAR-Indian Agricultural Statistics Research Institute, New Delhi, 110 012, India.
Mukesh Kumar
ICAR-Indian Agricultural Statistics Research Institute, New Delhi, 110 012, India.
Soumen Pal
ICAR-Indian Agricultural Statistics Research Institute, New Delhi, 110 012, India.
*Author to whom correspondence should be addressed.
Abstract
Wheat as an important cereal crop in India but presence of weeds results in significant damage in addition to insect pest and diseases. Weeds, which are unwanted plants that grow in agricultural crops, compete for essential elements like sunlight and water and are a major threat to food security. Conventional weed recognition approaches are very expensive, time consuming and require manual involvement by specialists. Researchers are actively investigating IT-based methods like computer vision and machine learning for weed identification. While models exist for identifying weeds in various crops, there is currently no specific model exists for weed identification in wheat crop. This paper proposed a mobile-based weed identification model using the ResNet50 deep learning architecture. The dataset used for training and testing the model consists of 1869 images of five prevalent weed species associated with wheat crop. After training, model demonstrated a notable accuracy of 93.25% on the validation dataset.
Keywords: Wheat, weed, CNN, Resnet50, mobile application