Efficient Mixture of Geographical Species for On Device Wildlife Monitoring
Not on the map. Why?
Abstract (arXiv)
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.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 33 of its 45 references that are in the map. First posted on arXiv 11 Apr 2025.
Builds on
33 of 45 references.
References · 33
- 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,795
- 2020Training data-efficient image transformers & distillation through attentioncited by 4,586
- 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,509
- 2020Supervised Contrastive Learningcited by 2,829
- 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,375
- 2021A Survey of Quantization Methods for Efficient Neural Network Inferencecited by 815
- 2018Recognition in Terra Incognitacited by 658
- 2017ThiNet: A Filter Level Pruning Method for Deep Neural Network Compressioncited by 442
- 2019BigEarthNet: A Large-Scale Benchmark Archive For Remote Sensing Image Understandingcited by 234
- 2020Geography-Aware Self-Supervised Learningcited by 190
- 2022Separable Self-attention for Mobile Vision Transformerscited by 163
- 2022Finetune like you pretrain: Improved finetuning of zero-shot vision modelscited by 150
21 more
- 2021Seeing biodiversity: perspectives in machine learning for wildlife conservationcited by 130
- 2020Federated Visual Classification with Real-World Data Distributioncited by 128
- 2016PlaNet - Photo Geolocation with Convolutional Neural Networkscited by 115
- 2018Tile2Vec: Unsupervised representation learning for spatially distributed datacited by 110
- 2021Accelerated Sparse Neural Training: A Provable and Efficient Method to Find N:M Transposable Maskscited by 103
- 2020Multi-Scale Representation Learning for Spatial Feature Distributions using Grid Cellscited by 90
- 2021The iWildCam 2021 Competition Datasetcited by 77
- 2022On the Importance and Applicability of Pre-Training for Federated Learningcited by 71
- 2023CSP: Self-Supervised Contrastive Spatial Pre-Training for Geospatial-Visual Representationscited by 69
- 2023FasterViT: Fast Vision Transformers with Hierarchical Attentioncited by 65
- 2023Spatial Implicit Neural Representations for Global-Scale Species Mappingcited by 59
- 2024A Survey on Transformer Compressioncited by 46
- 2019Geo-Aware Networks for Fine-Grained Recognitioncited by 44
- 2024TorchSpatial: A Location Encoding Framework and Benchmark for Spatial Representation Learningcited by 31
- 2019Similarity Learning Networks for Animal Individual Re-Identification -- Beyond the Capabilities of a Human Observercited by 18
- 2024LoRA-C: Parameter-Efficient Fine-Tuning of Robust CNN for IoT Devicescited by 10
- 2019Learning Geo-Temporal Image Featurescited by 9
- 2019Conservation AI: Live Stream Analysis for the Detection of Endangered Species Using Convolutional Neural Networks and Drone Technologycited by 5
- 2022Bag of Tricks for Long-Tail Visual Recognition of Animal Species in Camera-Trap Imagescited by 3
- 2023Vision Models Can Be Efficiently Specialized via Few-Shot Task-Aware Compressioncited by 2
- 2024Towards Vision Mixture of Experts for Wildlife Monitoring on the Edgecited by 2
Built on by
Not cited yet.