Accuracy and Efficiency Comparison of Object Detection Open-Source Models

In agriculture, weeds cause direct damage to the crop, and it primarily affects crop yield potential. Manual mechanical weeding methods consume a lot of energy time do not give efficient results. Chemical weed control is still best way weeds. However, widespread large-scale use herbicides harmful environment. Our study's objective propose an model for smart system detect in crops real-time using computer vision. experiment dataset contains images two different species well known our region strained this with temperate climate. The first Phalaris Paradoxa. second Convolvulus, manually captured professional camera from fields under lighting conditions (from morning afternoon sunny cloudy weather). detection has experimented four recent pre-configured open-source vision models object detection: Detectron2, EfficientDet, YOLO, Faster R-CNN. performance comparison executed on Open CV Keras platform python language.