A method for custom measurement of fish dimensions using the improved YOLOv5-keypoint framework with multi-attention mechanisms
Bibliographic record
Abstract
Dimensional data directly reflects the growth rate of individual fish, an important economic trait of interest to fish researchers. Efficiently obtaining large-scale fish dimension data would be valuable for both selective breeding and production. To address this, our study proposes a custom dimension measurement method for fish using the YOLOv5-keypoint framework with multi-attention mechanisms. We optimized the YOLOv5 framework, incorporated the SimAM attention mechanism to achieve more accurate and faster fish detection, and added customizable landmarks to the network structure, enabling flexible configuration of the number and location of feature points in the training dataset. This method is applicable to various aquacultural species and other objects. We tested the effectiveness of the method using the economically important grass carp ( Ctenopharyngodon idella ). The proposed method outperforms pure YOLOv5, Faster R-CNN, and SSD in terms of precision and recall rates, achieving an impressive average precision of 0.9781. Notably, field trials confirmed the method's exceptional measurement accuracy, exceeding 97% compatibility with manual measurements, while demonstrating a real-time speed of 38 frames per second on the NVIDIA RTX A4000. This enables efficient and accurate large-scale surface dimension measurements of economic fish. To facilitate massive measurements in agricultural research, we have implemented this method as an online platform, called Mode-recognition Ruler (MrRuler, http://bioinfo.ihb.ac.cn/mrruler ). The platform identifies objects in a single image at an average speed of 0.486 ± 0.005 s, based on a dataset of 10,000 images. MrRuler includes two preset carp models and allows users to upload training datasets for custom models of their targets of interest.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
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".