Crop Identification and Growth Stage Determination for Autonomous Navigation of Agricultural Robots

by Shivam Kashyap

With a focus toward mainly allowing mobile robot navigation, this work presents two methods for crop recognition and growth stage determination. These techniques comprise a two-phase approach with distinct models for crops and growth stage classification and a one-phase approach with a single model able of managing all crops and development phases. Using maize and sugar beet field images, the techniques were validated and shown to be both successful. For situations with a limited range of crops, the one-phase approach proved to be helpful since it allowed, with a single model, to identify both the type and growth state of the crop and shown an overall Mean Average Precision (mAP) of roughly 67.50%. Furthermore, the two-phase approach identified the crop type first, obtaining an overall mAP of roughly 74.2%, with maize detection showing quite good performance at 77.6%. Nevertheless, the mAP was only able to reach 61.3% when it came to determining the particular maize growth state since some challenges emerged when precisely classifying maize growth stages with six and eight leaves. Conversely, the two-phase approach is shown to be more scalable and flexible, hence it is a better option for systems allowing a variety of crops.

We’ve teamed up with sproutQ.com, one of India’s leading hiring platforms, to bring you a smarter, faster, and more personalized resume-building experience.

Leave a Reply

[script_17]

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. OK Read More

Privacy & Cookies Policy