Projekte pro Jahr
Abstract
Distribution shifts are characterized by differences between the training and test data distributions. They can significantly reduce the accuracy of machine learning models deployed in real-world scenarios. This paper explores the distribution shift problem when classifying pollen grains from microscopic images collected in the wild with a low-cost camera sensor. We leverage the domain knowledge that geometric features are highly important for accurate pollen identification and introduce two novel geometric image augmentation techniques to significantly narrow the accuracy gap between the model performance on the train and test datasets. In particular, we show that Tenengrad and ImageToSketch filters are highly effective to balance the shape and texture information while leaving out unimportant details that may confuse the model. Extensive evaluations on various model architectures demonstrate a consistent improvement of the model generalization to field data of up to 14% achieved by the geometric augmentation techniques when compared to a wide range of standard image augmentations. The approach is validated through an ablation study using pollen hydration tests to recover the shape of dry pollen grains. The proposed geometric augmentations also receive the highest scores according to the affinity and diversity measures from the literature.
Originalsprache | englisch |
---|---|
Seitenumfang | 9 |
Publikationsstatus | Veröffentlicht - 16 Dez. 2023 |
Veranstaltung | 29th IEEE International Conference on Parallel and Distributed Systems: IEEE ICPADS 2023 - Ocean Flower Island, Hainan, China Dauer: 17 Dez. 2023 → 21 Dez. 2023 https://ieee-cybermatics.org/2023/icpads/ |
Konferenz
Konferenz | 29th IEEE International Conference on Parallel and Distributed Systems |
---|---|
Kurztitel | IEEE ICPADS |
Land/Gebiet | China |
Ort | Hainan |
Zeitraum | 17/12/23 → 21/12/23 |
Internetadresse |
Fingerprint
Untersuchen Sie die Forschungsthemen von „Mitigating Distribution Shifts in Pollen Classification from Microscopic Images Using Geometric Data Augmentations“. Zusammen bilden sie einen einzigartigen Fingerprint.-
Intelligent & Networked Embedded Systems
Boano, C. A., Römer, K. U., Schuß, M., Cao, N., Saukh, O., Hofmann, R., Stocker, M., Schuh, M. P., Papst, F., Salomon, E., Brunner, H., Gallacher, M., Mohamed Hydher, M. H., Wang, D., Corti, F., Krisper, M., Basic, F. & Petrovic, K.
1/09/13 → 31/12/24
Projekt: Arbeitsgebiet
-
LocSense - Zuverlässige Lokalisierungs- und Erkennungsdienste für kognitive Produkte
Römer, K. U., Diwold, K., Cao, N. & Saukh, O.
15/12/20 → 31/03/21
Projekt: Forschungsprojekt
Publikationen
- 1 Preprint
-
Geometric Data Augmentations to Mitigate Distribution Shifts in Pollen Classification from Microscopic Images
Cao, N. & Saukh, O., 18 Nov. 2023, arXiv, 16 S.Publikation: Arbeitspapier › Preprint