ESC-YOLOv8-seg: A real-time non-destructive detection framework for small-target surface anomalies in zebrafish underwater monitoring
Bibliographic record
Abstract
The detection of anomalies on fish surfaces is of critical importance for assessing fish health status, preventing fish disease outbreaks, predicting changes in water quality, and enhancing fish welfare. The zebrafish ( Danio rerio ), a key model organism, has been increasingly utilized in various fields, including medicine, genetics and environmental toxicology. This has led to a corresponding increase in demand for intelligent management and detection systems. However, traditional methods of fish disease detection may have irreversible effects on fish, particularly small species, and often fail to meet the precision, non-destructive warning, and real-time requirements for zebrafish detection. To address this issue, this study proposes a novel method based on the YOLOv8 framework, designated ESC-YOLOv8-seg. This method significantly enhances the precision and speed of detecting surface abnormalities on small fish in complex settings by integrating the EMA, SPPELAN, and C2f-Faster modules, and incorporating an additional detection head (P2) optimized for the extreme small target size of zebrafish. Furthermore, the integration of positional information and surface features enables the method to achieve real-time monitoring and non-destructive early warning of fish surface abnormalities. The proposed method enhances precision in small target detection and achieves high accuracy in discerning subtle differences among detection targets. In real aquaculture settings, it can reach an average speed of 106 FPS with a detection accuracy of 98 %. Although this study has been designed to meet the specific needs of zebrafish scientific research, it is highly generalisable and can be applied to the real-time detection of underwater surface abnormalities in a range of fish species in aquaculture. • Provided a machine vision-based scheme for monitoring the health of zebrafish. • Proposed a novel method for fusing attention mechanisms with lightweight network. • Overcame challenges in segmentation and classification of small underwater objects. • Presented ESC-YOLOv8-seg, a novel network for target detection and segmentation. • ESC-YOLOv8-seg achieves 98 % accuracy at 108 FPS based on image enhancement.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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 source (direct Gemma or distilled Codex), 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".