Semi and Self-supervised Learning for Multi-label Image Classification: An Underwater Inspection Imagery Application
Bibliographic record
Abstract
Underwater inspections are essential for the preventive maintenance of offshore equipment but can be costly and time-consuming, requiring specialized professionals. In this scenario, we propose a solution that helps inspection specialists find key events/objects in video frames by applying image classification models. In order to validate this approach, we use frames of real inspections from an oil and gas company. Some of this application’s challenges are inherent to underwater image classification, such as uneven lighting conditions, low contrast, and distortion. In addition, we have two significant issues induced by human labeling: label scarcity and noise. Due to the complexity of the assignment, traditional supervised models might fulfill the task. We address this problem by applying DINO ("Self-DIstillation with NO labels") and a new multi-label version of PAWS ("Predicting view Assignments With Support samples"), mPAWS ("multi-label PAWS"). The results show the benefits of using such models, with an improvement in the F1-Score of 2.7% compared to state-of-the-art supervised models. The models can perform real-time inference once deployed in the finished product, which has the potential to accelerate the underwater inspection procedure, making the work of the specialists more time-efficient and less prone to human errors.
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 | 0.002 | 0.003 |
| 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 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 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".