Search Results
Optimized Deep Learning Model as a Basis for Fast UAV Mapping of Weed Species in Winter Wheat Crops
Weed maps should be available quickly, reliably, and with high detail to be useful for site-specific management in crop protection and to promote more sustainable agriculture by reducing pesticide use. Here, the optimization of a deep residual convolutional neural network (ResNet-18) for the classification of weed and crop plants in UAV imagery is proposed. The target was to reach sufficient performance on an embedded system by maintaining the same features of the ResNet-18 model as a basis for fast UAV mapping. This would enable online recognition and subsequent mapping of weeds during UAV flying operation. Optimization was achieved mainly by avoiding redundant computations that arise when a classification model is applied on overlapping tiles in a larger input image. The model was trained and tested with imagery obtained from a UAV flight campaign at low altitude over a winter wheat field, and classification was performed on species level with the weed species Matricaria chamomilla L., Papaver rhoeas L., Veronica hederifolia L., and Viola arvensis ssp. arvensis observed in that field. The ResNet-18 model with the optimized image-level prediction pipeline reached a performance of 2.2 frames per second with an NVIDIA Jetson AGX Xavier on the full resolution UAV image, which would amount to about 1.78 ha h−1 area output for continuous field mapping. The overall accuracy for determining crop, soil, and weed species was 94%. There were some limitations in the detection of species unknown to the model. When shifting from 16-bit to 32-bit model precision, no improvement in classification accuracy was observed, but a strong decline in speed performance, especially when a higher number of filters was used in the ResNet-18 model. Future work should be directed towards the integration of the mapping process on UAV platforms, guiding UAVs autonomously for mapping purpose, and ensuring the transferability of the models to other crop fields.
Regression kriging for improving crop height models fusing ultra-sonic sensing with UAV imagery
A crop height model (CHM) can be an important element of the decision making process in agriculture, because it relates well with many agronomic parameters, e.g., crop height, plant biomass or crop yield. Today, CHMs can be inexpensively obtained from overlapping imagery captured from unmanned aerial vehicle (UAV) platforms or from proximal sensors attached to ground-based vehicles used for regular management. Both approaches have their limitations and combining them with a data fusion may overcome some of these limitations. Therefore, the objective of this study was to investigate if regression kriging, as a geostatistical data fusion approach, can be used to improve the interpolation of ground-based ultrasonic measurements with UAV imagery as covariate. Regression kriging might be suitable because we have a sparse data set (ultrasound) and an exhaustive data set (UAV) and both data sets have favorable properties for geostatistical analysis. To confirm this, we conducted four missions in two different fields in total, where we collected UAV imagery and ultrasonic data alongside. From the overlapping UAV images, surface models and ortho-images were generated with photogrammetric processing. The maps generated by regression kriging were of much higher detail than the smooth maps generated by ordinary kriging, because regression kriging ensures that for each prediction point information from the UAV, imagery is given. The relationship with crop height, fresh biomass and, to a lesser extent, with crop yield, was stronger using CHMs generated by regression kriging than by ordinary kriging. The use of UAV data from the prior mission was also of benefit and could improve map accuracy and quality. Thus, regression kriging is a flexible approach for the integration of UAV imagery with ground-based sensor data, with benefits for precision agriculture-oriented farmers and agricultural service providers.