# Multi-Level Feature Fusion for Continual Learning in Visual Quality Inspection

arXiv: https://arxiv.org/abs/2601.00725 · HTML: https://mlsanity.com/paper/2601.00725/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 42 references that are in the map.
- First posted on arXiv 2 Jan 2026.

## Builds on (most cited first, 25 of 25)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [iCaRL: Incremental Classifier and Representation Learning](https://mlsanity.com/paper/1611.07725/) (2016)
- [FiLM: Visual Reasoning with a General Conditioning Layer](https://mlsanity.com/paper/1709.07871/) (2017)
- [Searching for MobileNetV3](https://mlsanity.com/paper/1905.02244/) (2019)
- [Gradient Episodic Memory for Continual Learning](https://mlsanity.com/paper/1706.08840/) (2017)
- [Progressive Neural Networks](https://mlsanity.com/paper/1606.04671/) (2016)
- [Continual Lifelong Learning with Neural Networks: A Review](https://mlsanity.com/paper/1802.07569/) (2018)
- [Swin Transformer V2: Scaling Up Capacity and Resolution](https://mlsanity.com/paper/2111.09883/) (2021)
- [Memory Aware Synapses: Learning what (not) to forget](https://mlsanity.com/paper/1711.09601/) (2017)
- [Dark Experience for General Continual Learning: a Strong, Simple Baseline](https://mlsanity.com/paper/2004.07211/) (2020)
- [A Comprehensive Survey of Continual Learning: Theory, Method and Application](https://mlsanity.com/paper/2302.00487/) (2023)
- [Vision Transformers Need Registers](https://mlsanity.com/paper/2309.16588/) (2023)
- [Online Continual Learning with Maximally Interfered Retrieval](https://mlsanity.com/paper/1908.04742/) (2019)
- [Re-evaluating Continual Learning Scenarios: A Categorization and Case for Strong Baselines](https://mlsanity.com/paper/1810.12488/) (2018)
- [Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processes](https://mlsanity.com/paper/1906.07697/) (2019)
- [Online Coreset Selection for Rehearsal-based Continual Learning](https://mlsanity.com/paper/2106.01085/) (2021)
- [Continual Learning with Foundation Models: An Empirical Study of Latent Replay](https://mlsanity.com/paper/2205.00329/) (2022)
- [Computationally Budgeted Continual Learning: What Does Matter?](https://mlsanity.com/paper/2303.11165/) (2023)
- [First Session Adaptation: A Strong Replay-Free Baseline for Class-Incremental Learning](https://mlsanity.com/paper/2303.13199/) (2023)
- [Transfer without Forgetting](https://mlsanity.com/paper/2206.00388/) (2022)
- [Efficient Continual Learning in Neural Networks with Embedding Regularization](https://mlsanity.com/paper/1909.03742/) (2019)
- [GRASP: A Rehearsal Policy for Efficient Online Continual Learning](https://mlsanity.com/paper/2308.13646/) (2023)
- [DLCFT: Deep Linear Continual Fine-Tuning for General Incremental Learning](https://mlsanity.com/paper/2208.08112/) (2022)
- [A Dataset and Baseline for Deep Learning-Based Visual Quality Inspection in Remanufacturing](https://mlsanity.com/paper/2511.15440/) (2025)

## Abstract (arXiv)

Deep neural networks show great potential for automating various visual quality inspection tasks in manufacturing. However, their applicability is limited in more volatile scenarios, such as remanufacturing, where the inspected products and defect patterns often change. In such settings, deployed models require frequent adaptation to novel conditions, effectively posing a continual learning problem. To enable quick adaptation, the necessary training processes must be computationally efficient while still avoiding effects like catastrophic forgetting. This work presents a multi-level feature fusion (MLFF) approach that aims to improve both aspects simultaneously by utilizing representations from different depths of a pretrained network. We show that our approach is able to match the performance of end-to-end training for different quality inspection problems while using significantly less trainable parameters. Furthermore, it reduces catastrophic forgetting and improves generalization robustness to new product types or defects.