A Comparative Evaluation of Vision Language Models for Waste Classification in Few-Shot Settings
Bibliographic record
Abstract
Efficient waste classification is essential for sustainable waste management systems.Accurate sorting can significantly enhance recycling efforts and reduce pollution.However, traditional computer vision methods often require large, annotated datasets and extensive retraining, limiting their adaptability to varying waste types and challenging real-world conditions.In this study, we evaluate the potential of Multimodal Large Language Models (MLLMs) and Vision-Language Models (VLMs) for adaptive waste classification, focusing on zero-shot and few-shot learning scenarios.Using datasets such as TrashNet and our custom MultiWaste dataset, we test a method using a CLIP VLM for feature extraction and a simple Nearest Neighbour (VLM-NN) approach for classification.This showcases robust few-shot capabilities and excellent scalability, achieving an accuracy of 97.74% on TrashNet.While MLLMs exhibit strong zeroshot capabilities, their utility diminishes with increasing labelled samples due to high computational costs.In contrast, VLM-NN offers efficient performance but struggles with extremely limited training data.Our results show the potential of Large Pretrained Models for the task of waste classification while providing guidance on which model architectures to consider for different amounts of training data.
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 | 0.004 | 0.010 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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".