Algorithmic geolocation of harvest in hand-picked agriculture

Abstract: Precision agriculture significantly depends on measuring yield; this allows feedback to optimize various decisions. While spatially granular yield mapping is readily available in machine-harvested row crops, it is more difficult in hand-picked row crops. We study here a data set collected during harvesting of strawberries; using smartphones, we collected Global Positioning System (GPS) logs of individual harvesters. Using recent advances in feature identification, we are able to algorithmically decompose the path into individual excursions into the field to harvest the berries. This lays the groundwork for yield mapping. To further develop this area, we recommend that Resource Managers: Pursue wider scale trials of geolocated harvest data collection of hand-picked crops. Join this geolocated harvest data with data from other aspects of field operations. Join this geolocated harvest data with output measurements like quality and quantity.