Re-identification for long-term tracking and management of health and welfare challenges in pigs
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
The fine-grained management of farm animals using video analytics relies on long-term visual tracking of individual animals. When an individual is occluded or exits the camera's field of view, tracking can be lost. The problem of using visual features to re-assign identity after loss of tracking is known as re-identification. In the case of pigs , this problem is especially challenging due to the similar appearances of most individual animals within a pen. To address this issue, an image-based pig re-identification method is developed that is invariant to pose, illumination, and camera viewpoint. This method allows pigs to be reidentified, enabling long-term monitoring. This approach uses a Vision-Transformer model (ViT) previously developed for person re-identification. The model was trained using specifically designed pig re-identification datasets with a diverse range of housing and management conditions. These datasets use overhead cameras, allowing an investigation of the effect of the detection approach on re-identification performance. Re-identification using a traditional axis-aligned pig detector was compared with a recently developed oriented pig detector that better matches the pig's pose when extracting the pig from the wider image. It was found that the use of an oriented detector led to improved performance. The proposed system achieved a peak rank-1 Cumulative Matching Characteristic (CMC) performance of 90.5%. Furthermore, it is shown that this model is capable of generalising across different farms, achieving an average rank-1 CMC 81.8% in the cross-farm setting. Finally, the proposed re-identification features can be incorporated into an existing multi-object tracking system to improve its performance at reacquiring pig identities when tracking is lost. Overall, this work demonstrates the potential of using visual re-identification features of pigs to enable individual-level animal management.
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.
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 | 0.000 | 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 it