A Deep Learning-Based Novel Approach for Weed Growth Estimation

Automation of agricultural food production is growing in popularity scientific communities and industry. The main goal automation to identify detect weeds the crop. Weed intervention for duration crop establishment a serious difficulty wheat North India. soil nutrient important production. Weeds usually compete light, water air nutrients space from target This research paper assesses growth rate due macronutrients (nitrogen, phosphorus potassium) absorbed various soils (fertile, clay loamy) rabi field. weed image data have been collected three different places Madhya Pradesh, India with 10 crops (Maize, Lucerne, Cumin, Coriander, Wheat, Fenugreek, Gram, Onion, Mustard Tomato) (Corchorus Capsularis, Cynodondactylon, Chloris barbata, Amaranthaceae, Argemone mexicana, Carthamus oxyacantha, Capsella bursa Pastoris, Chenopodium Album, Dactyloctenium aegyptium Convolvulus Ravens). Intel Real Sense LiDAR digital camera L515 Canon SLR DIGICAM EOS 850 D 18-55IS STM cameras were mounted over × square feet area land 3670 images collected. 2936 used training 734 testing validation. Efficient Net-B7 Inception V4 architectures train model that has provided accuracy 97% 94% respectively. Image classification using Inspection was unsuccessful less accurate results as compared EfficientNet-B7.