Data for: Producing Plankton Classifiers that are Robust to Dataset Shift

Modern plankton high-throughput monitoring relies on deep learning classifiers for species recognition in water ecosystems. Despite satisfactory nominal performances, a significant challenge arises from the dataset shift, where performance drops during real-world deployment compared to ideal testing conditions. In our study, we integrate the ZooLake dataset, which consists of dark-field images of lake plankton, with manually-annotated images from 10 independent days of deployment, serving as test cells to benchmark out-of-dataset (OOD) performances. Our analysis reveals instances where classifiers, initially performing well in ideal conditions, encounter notable failures in real-world scenarios. For example, a MobileNet with a 92% nominal test accuracy shows a 77% OOD accuracy. We systematically investigate conditions leading to OOD performance drops and propose a preemptive assessment method to identify potential pitfalls when classifying new data, and pinpoint features in OOD images that adversely impact classification. We present a three-step pipeline: (i) identifying OOD degradation compared to nominal test performance, (ii) conducting a diagnostic analysis of degradation causes, and (iii) providing solutions. We find that ensembles of BEiT vision transformers, with targeted augmentations addressing OOD robustness, geometric ensembling, and rotation-based test-time augmentation, constitute the most robust model. It achieves an 83% OOD accuracy, with errors concentrated on container classes. Moreover, it exhibits lower sensitivity to dataset shift, and reproduces well the plankton abundances. Our proposed pipeline is applicable to generic plankton classifiers, contingent on the availability of suitable test cells. Implementation of this pipeline is anticipated to usher in a new era of robust classifiers, resilient to dataset shift, and capable of delivering reliable plankton abundance data. By identifying critical shortcomings and offering practical procedures to fortify models against dataset shift, our study contributes to the development of more reliable plankton classification technologies.