# Half a Percent of Labels is Enough: Efficient Animal Detection in UAV Imagery using Deep CNNs and Active Learning

arXiv: https://arxiv.org/abs/1907.07319 · HTML: https://mlsanity.com/paper/1907.07319/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Jul 2019.

## Built on by (most cited first, 22 of 22)

- [Extending the WILDS Benchmark for Unsupervised Adaptation](https://mlsanity.com/paper/2112.05090/) (2021)
- [On Statistical Bias In Active Learning: How and When To Fix It](https://mlsanity.com/paper/2101.11665/) (2021)
- [DIAL: Deep Interactive and Active Learning for Semantic Segmentation in Remote Sensing](https://mlsanity.com/paper/2201.01047/) (2022)
- [Mapping oil palm density at country scale: An active learning approach](https://mlsanity.com/paper/2105.11207/) (2021)
- [Automated wildlife image classification: An active learning tool for ecological applications](https://mlsanity.com/paper/2303.15823/) (2023)
- [Active Learning for Improved Semi-Supervised Semantic Segmentation in Satellite Images](https://mlsanity.com/paper/2110.07782/) (2021)
- [An Informative Path Planning Framework for Active Learning in UAV-based Semantic Mapping](https://mlsanity.com/paper/2302.03347/) (2023)
- [Self-Supervised Pretraining and Controlled Augmentation Improve Rare Wildlife Recognition in UAV Images](https://mlsanity.com/paper/2108.07582/) (2021)
- [Relating tSNE and UMAP to Classical Dimensionality Reduction](https://mlsanity.com/paper/2306.11898/) (2023)
- [Automated Detection of Cat Facial Landmarks](https://mlsanity.com/paper/2310.09793/) (2023)
- [Find Rhinos without Finding Rhinos: Active Learning with Multimodal Imagery of South African Rhino Habitats](https://mlsanity.com/paper/2409.18104/) (2024)
- [SAR Strikes Back: A New Hope for RSVQA](https://mlsanity.com/paper/2501.08131/) (2025)
- [Small or Far Away? Exploiting Deep Super-Resolution and Altitude Data for Aerial Animal Surveillance](https://mlsanity.com/paper/2111.06830/) (2021)
- [Mapping on a Budget: Optimizing Spatial Data Collection for ML](https://mlsanity.com/paper/2509.03749/) (2025)
- [Rare Wildlife Recognition with Self-Supervised Representation Learning](https://mlsanity.com/paper/2211.05636/) (2022)
- [Understanding Approximation for Bayesian Inference in Neural Networks](https://mlsanity.com/paper/2211.06139/) (2022)
- [SharkTrack: an accurate, generalisable software for streamlining shark and ray underwater video analysis](https://mlsanity.com/paper/2407.20623/) (2024)
- [Oh Deer, How Should I Handle This? Seasonal Priors for Selective Wildlife Annotation and Classification](https://mlsanity.com/paper/2608.02762/) (2026)
- [Deep Learning and Earth Observation to Support the Sustainable Development Goals](https://mlsanity.com/paper/2112.11367/) (2021)
- [CubeletWorld: A New Abstraction for Scalable 3D Modeling](https://mlsanity.com/paper/2511.17664/) (2025)
- [iSAGE: A Human-in-the-Loop Framework for Remote Sensing Semantic Segmentation via Sparse Point Supervision](https://mlsanity.com/paper/2606.10136/) (2026)
- [When One Modality Is Not Enough: Multimodal Sex and Life-Stage Classification of Red Deer from Aerial RGB-Thermal Video](https://mlsanity.com/paper/2608.06973/) (2026)

## Abstract (arXiv)

We present an Active Learning (AL) strategy for re-using a deep Convolutional Neural Network (CNN)-based object detector on a new dataset. This is of particular interest for wildlife conservation: given a set of images acquired with an Unmanned Aerial Vehicle (UAV) and manually labeled gound truth, our goal is to train an animal detector that can be re-used for repeated acquisitions, e.g. in follow-up years. Domain shifts between datasets typically prevent such a direct model application. We thus propose to bridge this gap using AL and introduce a new criterion called Transfer Sampling (TS). TS uses Optimal Transport to find corresponding regions between the source and the target datasets in the space of CNN activations. The CNN scores in the source dataset are used to rank the samples according to their likelihood of being animals, and this ranking is transferred to the target dataset. Unlike conventional AL criteria that exploit model uncertainty, TS focuses on very confident samples, thus allowing a quick retrieval of true positives in the target dataset, where positives are typically extremely rare and difficult to find by visual inspection. We extend TS with a new window cropping strategy that further accelerates sample retrieval. Our experiments show that with both strategies combined, less than half a percent of oracle-provided labels are enough to find almost 80% of the animals in challenging sets of UAV images, beating all baselines by a margin.