Monitoring crop phenology with street-level imagery using computer vision

Abstract

• Method to automatically collect side looking imagery for agriculture parcels. • Crop types and phenology obtained from computer vision. • Demonstration over monthly surveyed totalling 400,000 pictures for 17 crop types. • Deep learning hyper-parameterization for street view imagery for crop and phenology. • Parcel discrimination has M-F1 of 88.1% for crops and 86.9% for phenological stages. Street-level imagery holds a significant potential to scale-up in-situ data collection. This is enabled by combining the use of cheap high-quality cameras with recent advances in deep learning compute solutions to derive relevant thematic information. We present a framework to collect and extract crop type and phenological information from street level imagery using computer vision. Monitoring crop phenology is critical to assess gross primary productivity and crop yield. During the 2018 growing season, high-definition pictures were captured with side-looking action cameras in the Flevoland province of the Netherlands. Each month from March to October, a fixed 200-km route was surveyed collecting one picture per second resulting in a total of 400,000 geo-tagged pictures. At 220 specific parcel locations, detailed on the spot crop phenology observations were recorded for 17 crop types (including bare soil, green manure, and tulips): bare soil, carrots, green manure, grassland, grass seeds, maize, onion, potato, summer barley, sugar beet, spring cereals, spring wheat, tulips, vegetables, winter barley, winter cereals and winter wheat. Furthermore, the time span included specific pre-emergence parcel stages, such as differently cultivated bare soil for spring and summer crops as well as post-harvest cultivation practices, e.g. green manuring and catch crops. Classification was done using TensorFlow with a well-known image recognition model, based on transfer learning with convolutional neural network (MobileNet). A hypertuning methodology was developed to obtain the best performing model among 160 models. This best model was applied on an independent inference set discriminating crop type with a Macro F1 score of 88.1% and main phenological stage at 86.9% at the parcel level. Potential and caveats of the approach along with practical considerations for implementation and improvement are discussed. The proposed framework speeds up high quality in-situ data collection and suggests avenues for massive data collection via automated classification using computer vision.