On the Robustness and Real-time Adaptation of UAV-based Crowd Localization in Complex and High-density Scenes
Bibliographic record
Abstract
This paper presents a novel approach for robust and real-time crowd localization using unmanned aerial vehicles (UAVs) in complex scenarios. The approach follows the single-stage detection paradigm, aiming to balance time and space complexity efficiently. To achieve this, the proposed method leverages pertaining of the architecture with UAV-oriented object detection datasets, specifically VisDrone and UAVDT. Additionally, selective data augmentations are applied to adapt the model to the crowd localization vision task in the UAV-oriented theme, using datasets such as NWPU-Crowd, UCF-QNRF, and Shanghai Tech. The authors conducted ablation studies to assess the contributions of pretraining and different adaptation techniques to the overall performance. The results demonstrate that the proposed approach achieves high performance on the edge, specifically on the NVIDIA Jetson Xavier NX board, with well-balanced options that align with the application design requirements. This enables effective crowd localization in complex scenarios using UAVs. Furthermore, the study sheds light on the lack of existing datasets specifically designed for UAV-based crowd localization and proposes potential solutions to address this gap. By presenting both quantitative and qualitative results, the paper shows the effectiveness of the proposed approach, achieving satisfactory performance while reducing inference time and the number of parameters. This work has good potential to significantly contribute to the robotics and automation society by introducing the first approach for UAV-based crowd localization on the edge platform. It also provides valuable insights into the importance of pretraining and data adaptations techniques in this context. Overall, the proposed approach offers a promising solution for crowd localization in complex scenarios using UAVs, and it has the potential to advance various applications in real-world scenarios.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".