Mikko Impiö
I am a researcher and PhD candidate at the Finnish Environment Institute (Syke), working on deep learning and computer vision for or environmental monitoring.
My work focuses on topics like fine-grained classification, out-of-distribution detection and multimodal learning from images, videos and DNA data, with applications in biodiversity monitoring and remote sensing.
I hold a MSc in Electrical Engineering from Tampere University, with a major in Signal Processing. During my studies I completed internships at Intel, where I worked with 3A algorithms, and at Vaisala, focusing on deep learning methods for predictive maintenance.
I am a member in several expert groups, including:
- EIONET Data and Digitalization (European Environment Agency).
- Lukki, Finland’s national coordination group for nature information.
- GEO AI4EO, a GEO subgroup focused on geospatial AI for Earth Observation.
Blog
Publications
Impiö, M., Raitoharju, J. (2025). Efficient curation of invertebrate image datasets using feature embeddings and automatic size comparison. In 2025 IEEE Symposia on Computational Intelligence for Energy, Transport and Environmental Sustainability (CIETES). IEEE. Paper | Code
Impiö, M., Raitoharju, J. (2024). Improving taxonomic image-based out-of-distribution detection with DNA barcodes. In 2024 32nd European Signal Processing Conference (EUSIPCO). IEEE. Paper
Impiö, M., Härmä, P., Tammilehto, A., Anttila, S., & Raitoharju, J. (2022). Habitat classification from satellite observations with sparse annotations. arXiv preprint arXiv:2209.12995. Paper
Impiö, M., Yamaç, M., & Raitoharju, J. (2021, June). Multi-level reversible encryption for ECG signals using compressive sensing. In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 1005-1009). IEEE. Paper | Code
de Schaetzen, F., Impiö, M., Wagner, B., Nienaltowski, P., Arnold, M., Huber, M., … & Stocker, R. (2023). The Riverine Organism Drift Imager: A new technology to study organism drift in rivers and streams. Methods in Ecology and Evolution. Paper
Code
dinotool
Command-line tool for extracting DINO features for images and videos
taxonomist
A library for training deep learning models for species classification.
tiers
A hierarchical label handling library for Python
point-eo
A python libary that makes it simple to sample points from large rasters, fit ML models and perform inference on larger-than-memory rasters.