Semi and Self-supervised Learning for Multi-label Classification for an Underwater Inspection Imagery Application
Bibliographic record
Abstract
<title>Abstract</title> Underwater inspections are crucial for the preventive maintenance of offshore equipment from the oil and gas industry. However, the entire inspection process is costly, subjective and time-consuming. Traditionally, specialists assess equipment conditions through image and sensor data collected by underwater vehicles that travel to the seabed. Since this data can present a wide range of events, an inspection can yield a considerable amount of data. Despite being a tedious task, this data is undoubtedly domain-specific, requiring a group of highly specialized professionals to correctly scrutinize it. In this scenario, we propose using image classification models to help specialists find event(s) of interest. However, there are challenges inherent to the underwater inspection image classification problem, such as: high cost and scarcity of balanced labeled data, presence of label noise, high intra-class variance and characteristics of underwater images such as turbidity and uneven lightning conditions. Therefore, traditional supervised models might not fulfill the task. We undertake this problem with the methods DINO (Self-DIstillation with NO labels, self-supervised) and a new proposed method mPAWS, a multi-label version of PAWS (Predicting View Assignments With Support Samples, semi-supervised). The results obtained show the benefits of using such models for the application, achieving an improvement of $2.7\%$ when compared to current state-of-the-art supervised models. Such models can perform real-time inference, which can help accelerate the tasks of the underwater vehicle, making the work of the specialists more time-efficient.
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 | 0.001 | 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 | 0.001 |
| 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".