Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2504.08620 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913788847980544 |
|---|---|
| author | Mensah, Emmanuel Azuh Mand, Joban Ou, Yueheng Jang, Min Heimerl, Kurtis |
| author_facet | Mensah, Emmanuel Azuh Mand, Joban Ou, Yueheng Jang, Min Heimerl, Kurtis |
| contents | Efficient on-device models have become attractive for near-sensor insight generation, of particular interest to the ecological conservation community. For this reason, deep learning researchers are proposing more approaches to develop lower compute models. However, since vision transformers are very new to the edge use case, there are still unexplored approaches, most notably conditional execution of subnetworks based on input data. In this work, we explore the training of a single species detector which uses conditional computation to bias structured sub networks in a geographically-aware manner. We propose a method for pruning the expert model per location and demonstrate conditional computation performance on two geographically distributed datasets: iNaturalist and iWildcam. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_08620 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Efficient Mixture of Geographical Species for On Device Wildlife Monitoring Mensah, Emmanuel Azuh Mand, Joban Ou, Yueheng Jang, Min Heimerl, Kurtis Computer Vision and Pattern Recognition Efficient on-device models have become attractive for near-sensor insight generation, of particular interest to the ecological conservation community. For this reason, deep learning researchers are proposing more approaches to develop lower compute models. However, since vision transformers are very new to the edge use case, there are still unexplored approaches, most notably conditional execution of subnetworks based on input data. In this work, we explore the training of a single species detector which uses conditional computation to bias structured sub networks in a geographically-aware manner. We propose a method for pruning the expert model per location and demonstrate conditional computation performance on two geographically distributed datasets: iNaturalist and iWildcam. |
| title | Efficient Mixture of Geographical Species for On Device Wildlife Monitoring |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2504.08620 |