Enhancing Drowning Surveillance with a Hybrid Vision Transformer Model: A Deep Learning Approach
Bibliographic record
Abstract
Annually, drowning claims the lives of approximately 372,200 individuals worldwide, averaging 40 fatalities per hour.In response, various technological advancements have been explored, including deep learning-based video and image processing, and wearable devices integrated with human pulse sensors and Light emitting diode (LED)/Liquid-crystal display (LCD) technologies.Despite these efforts, existing solutions have yet to fully address the challenge of accurate drowning detection.This study introduces a novel approach, leveraging a hybrid model that combines a traditional Vision Transformer (ViT) with plain Convolutional Neural Networks (CNNs).This model demonstrates a notable accuracy of 91.5% on a specialized dataset comprising 14,736 images of swimming and drowning scenarios, surpassing conventional methods in efficiency and size.In contrast to larger models like Swin-B, which comprises 88M parameters and achieves a marginally higher accuracy of 92.3%, the proposed model maintains high performance with only 5.9M parameters.The model's development involved pre-training on the ImageNet1K dataset, followed by fine-tuning using the specifically curated local dataset.The resultant system offers a cost-effective, efficient, and compact solution for drowning detection, suitable for various applications.This advancement in drowning surveillance technology highlights the potential of integrating ViT with CNNs in creating effective, resource-efficient models for critical real-world applications.
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 | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 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".