Mexican poppy (Argemone mexicana) control in cornfield using deep learning neural networks: a perspective

Abstract

ABSTRACTMexican poppy (Argemone mexicana) is a widespread noxious annual weed associated with crops such as corn (Zea mays L.), and this weed is persistent because it produces a seed bank. This invasive weed species must be controlled even in the dry season because Mexican poppy has a deep-reaching root system, which taps water from deep soil layers. Cases of a human death caused by Mexican poppy seeds in South Africa, India, and other Eastern countries were reported from the early years of the twentieth century. However, when weeds are controlled uniformly instead of site-specific or precision farming method across the spatially variable fields, there are environmental pollution challenges. Site-specific weed control techniques have gained interest in the precision farming community over the last years mainly because of Global Positioning System (GPS) applications, and a controlled measure of herbicides are applied where there are weeds in the field, and areas with more clusters of weeds receive the correct amount of herbicide application. Mexican poppy has prickles and is a nuisance to farmers, and herbicides represent a severe health hazard to humans due to chemical concentrations in water. For that reason, we propose the design of a site-specific weed control plan to use a row-guided robot to detect and identify weeds with accuracy, control speed timeously, and spray herbicides with a high level of precision and automation. These robotics methods are reported to be environmentally conscious, and economically efficient with less labour and management. The proposed method of deep learning neural networks, which use row-guided robots, a machine is trained on multiple images to identify weeds automatically from the main crop, and release a controlled measure of herbicides based on weed location and density, and spray weeds on-the-go upon emergence.