Search papers, labs, and topics across Lattice.
This paper investigates domain adaptation techniques for Rumex obtusifolius detection, training models on ground vehicle imagery and testing on drone-acquired imagery. CNNs struggled with this domain shift, but domain adaptation techniques like moment matching and maximum classifier discrepancy improved performance. Critically, Vision Transformers pretrained with self-supervision (DINOv2/3) exhibited strong generalization to the target domain, outperforming even adapted CNNs, achieving F1 scores of 0.8 after fine-tuning.
Self-supervised Vision Transformers can handily outperform domain-adapted CNNs when transferring weed detection models from ground-based to drone-based imagery.
Domain adaptation (DA) addresses the challenge of transferring a machine learning model trained on a source domain to a target domain with a different data distribution. In this work, we study DA for the task of Rumex obtusifolius (Rumex) image classification. We train models on a published, ground vehicle-based dataset (source) and evaluate their performance on a custom target dataset acquired by unmanned aerial vehicles (UAVs). We find that Convolutional Neural Network (CNN) models, specifically ResNets, generalize poorly to the target domain, even after fine-tuning on the source data. Applying moment-matching and maximum classifier discrepancy, two established DA techniques, substantially improves target-domain performance. However, Vision Transformer (ViT) models pretrained with self-supervised objectives (DINOv2, DINOv3) handle domain shifts intrinsically well, surpassing even moment-matching-trained ResNets, likely due to the rich, general-purpose representations acquired during large-scale pretraining. Using ViTs fine-tuned on the source dataset, we demonstrate high classification performances in the range of F1=0.8 on our target dataset. To support further research on DA for weed detection in grassland systems, we publicly release our UAV-based target dataset AGSMultiRumex, comprising data from 15 flights over Swiss meadows.
