Zijun Long

Zijun Long, George Killick, Richard McCreadie et al.

Robotic vision applications often necessitate a wide range of visual perception tasks, such as object detection, segmentation, and identification. While there have been substantial advances in these individual tasks, integrating specialized models into a unified vision pipeline presents significant engineering challenges and costs. Recently, Multimodal Large Language Models (MLLMs) have emerged as novel backbones for various downstream tasks. We argue that leveraging the pre-training capabilities of MLLMs enables the creation of a simplified framework, thus mitigating the need for task-specific encoders. Specifically, the large-scale pretrained knowledge in MLLMs allows for easier fine-tuning to downstream robotic vision tasks and yields superior performance. We introduce the RoboLLM framework, equipped with a BEiT-3 backbone, to address all visual perception tasks in the ARMBench challenge-a large-scale robotic manipulation dataset about real-world warehouse scenarios. RoboLLM not only outperforms existing baselines but also substantially reduces the engineering burden associated with model selection and tuning. The source code is publicly available at https://github.com/longkukuhi/armbench.

Zijun Long, Zaiqiao Meng, Gerardo Aragon Camarasa et al.

Vision Transformers (ViTs) have emerged as popular models in computer vision, demonstrating state-of-the-art performance across various tasks. This success typically follows a two-stage strategy involving pre-training on large-scale datasets using self-supervised signals, such as masked random patches, followed by fine-tuning on task-specific labeled datasets with cross-entropy loss. However, this reliance on cross-entropy loss has been identified as a limiting factor in ViTs, affecting their generalization and transferability to downstream tasks. Addressing this critical challenge, we introduce a novel Label-aware Contrastive Training framework, LaCViT, which significantly enhances the quality of embeddings in ViTs. LaCViT not only addresses the limitations of cross-entropy loss but also facilitates more effective transfer learning across diverse image classification tasks. Our comprehensive experiments on eight standard image classification datasets reveal that LaCViT statistically significantly enhances the performance of three evaluated ViTs by up-to 10.78% under Top-1 Accuracy.

Zijun Long, George Killick, Lipeng Zhuang et al.

Image classification datasets exhibit a non-negligible fraction of mislabeled examples, often due to human error when one class superficially resembles another. This issue poses challenges in supervised contrastive learning (SCL), where the goal is to cluster together data points of the same class in the embedding space while distancing those of disparate classes. While such methods outperform those based on cross-entropy, they are not immune to labeling errors. However, while the detrimental effects of noisy labels in supervised learning are well-researched, their influence on SCL remains largely unexplored. Hence, we analyse the effect of label errors and examine how they disrupt the SCL algorithm's ability to distinguish between positive and negative sample pairs. Our analysis reveals that human labeling errors manifest as easy positive samples in around 99% of cases. We, therefore, propose D-SCL, a novel Debiased Supervised Contrastive Learning objective designed to mitigate the bias introduced by labeling errors. We demonstrate that D-SCL consistently outperforms state-of-the-art techniques for representation learning across diverse vision benchmarks, offering improved robustness to label errors.

Zijun Long, George Killick, Richard McCreadie et al.

State-of-the-art image models predominantly follow a two-stage strategy: pre-training on large datasets and fine-tuning with cross-entropy loss. Many studies have shown that using cross-entropy can result in sub-optimal generalisation and stability. While the supervised contrastive loss addresses some limitations of cross-entropy loss by focusing on intra-class similarities and inter-class differences, it neglects the importance of hard negative mining. We propose that models will benefit from performance improvement by weighting negative samples based on their dissimilarity to positive counterparts. In this paper, we introduce a new supervised contrastive learning objective, SCHaNe, which incorporates hard negative sampling during the fine-tuning phase. Without requiring specialized architectures, additional data, or extra computational resources, experimental results indicate that SCHaNe outperforms the strong baseline BEiT-3 in Top-1 accuracy across various benchmarks, with significant gains of up to $3.32\%$ in few-shot learning settings and $3.41\%$ in full dataset fine-tuning. Importantly, our proposed objective sets a new state-of-the-art for base models on ImageNet-1k, achieving an 86.14\% accuracy. Furthermore, we demonstrate that the proposed objective yields better embeddings and explains the improved effectiveness observed in our experiments.

Xingwu Zhang, Guanxuan Li, Paul Henderson et al.

Current state-of-the-art multi-class unsupervised anomaly detection (MUAD) methods rely on training encoder-decoder models to reconstruct anomaly-free features. We first show these approaches have an inherent fidelity-stability dilemma in how they detect anomalies via reconstruction residuals. We then abandon the reconstruction paradigm entirely and propose Retrieval-based Anomaly Detection (RAD). RAD is a training-free approach that stores anomaly-free features in a memory and detects anomalies through multi-level retrieval, matching test patches against the memory. Experiments demonstrate that RAD achieves state-of-the-art performance across four established benchmarks (MVTec-AD, VisA, Real-IAD, 3D-ADAM) under both standard and few-shot settings. On MVTec-AD, RAD reaches 96.7\% Pixel AUROC with just a single anomaly-free image compared to 98.5\% of RAD's full-data performance. We further prove that retrieval-based scores theoretically upper-bound reconstruction-residual scores. Collectively, these findings overturn the assumption that MUAD requires task-specific training, showing that state-of-the-art anomaly detection is feasible with memory-based retrieval. Our code is available at https://github.com/longkukuhi/RAD.

Zijun Long, Xuri Ge, Richard Mccreadie et al.

Text-to-image retrieval aims to find the relevant images based on a text query, which is important in various use-cases, such as digital libraries, e-commerce, and multimedia databases. Although Multimodal Large Language Models (MLLMs) demonstrate state-of-the-art performance, they exhibit limitations in handling large-scale, diverse, and ambiguous real-world needs of retrieval, due to the computation cost and the injective embeddings they produce. This paper presents a two-stage Coarse-to-Fine Index-shared Retrieval (CFIR) framework, designed for fast and effective large-scale long-text to image retrieval. The first stage, Entity-based Ranking (ER), adapts to long-text query ambiguity by employing a multiple-queries-to-multiple-targets paradigm, facilitating candidate filtering for the next stage. The second stage, Summary-based Re-ranking (SR), refines these rankings using summarized queries. We also propose a specialized Decoupling-BEiT-3 encoder, optimized for handling ambiguous user needs and both stages, which also enhances computational efficiency through vector-based similarity inference. Evaluation on the AToMiC dataset reveals that CFIR surpasses existing MLLMs by up to 11.06% in Recall@1000, while reducing training and retrieval times by 68.75% and 99.79%, respectively. We will release our code to facilitate future research at https://github.com/longkukuhi/CFIR.

Zhuocheng Zhang, Xingwu Zhang, Kangheng Liang et al.

Recent advances in interactive text-to-image retrieval (I-TIR) use diffusion models to bridge the modality gap between the textual information need and the images to be searched, resulting in increased effectiveness. However, existing frameworks fuse multi-modal views of user feedback by simple embedding addition. In this work, we show that this static and undifferentiated fusion indiscriminately incorporates generative noise produced by the diffusion model, leading to performance degradation for up to 55.62% samples. We further propose ADaFuSE (Adaptive Diffusion-Text Fusion with Semantic-aware Experts), a lightweight fusion model designed to align and calibrate multi-modal views for diffusion-augmented I-TIR, which can be plugged into existing frameworks without modifying the backbone encoder. Specifically, we introduce a dual-branch fusion mechanism that employs an adaptive gating branch to dynamically balance modality reliability, alongside a semantic-aware mixture-of-experts branch to capture fine-grained cross-modal nuances. Via thorough evaluation over four standard I-TIR benchmarks, ADaFuSE achieves state-of-the-art performance, surpassing DAR by up to 3.49% in Hits@10 with only a 5.29% parameter increase, while exhibiting stronger robustness to noisy and longer interactive queries. These results show that generative augmentation coupled with principled fusion provides a simple, generalizable alternative to fine-tuning for interactive retrieval.

Xingwu Zhang, Guanxuan Li, Zhuocheng Zhang et al.

The rapidly growing number of product categories in large-scale e-commerce makes accurate object identification for automated packing in warehouses substantially more difficult. As the catalog grows, intra-class variability and a long tail of rare or visually similar items increase, and when combined with diverse packaging, cluttered containers, frequent occlusion, and large viewpoint changes-these factors amplify discrepancies between query and reference images, causing sharp performance drops for methods that rely solely on 2D appearance features. Thus, we propose RoboEye, a two-stage identification framework that dynamically augments 2D semantic features with domain-adapted 3D reasoning and lightweight adapters to bridge training deployment gaps. In the first stage, we train a large vision model to extract 2D features for generating candidate rankings. A lightweight 3D-feature-awareness module then estimates 3D feature quality and predicts whether 3D re-ranking is necessary, preventing performance degradation and avoiding unnecessary computation. When invoked, the second stage uses our robot 3D retrieval transformer, comprising a 3D feature extractor that produces geometry-aware dense features and a keypoint-based matcher that computes keypoint-correspondence confidences between query and reference images instead of conventional cosine-similarity scoring. Experiments show that RoboEye improves Recall@1 by 7.1% over the prior state of the art (RoboLLM). Moreover, RoboEye operates using only RGB images, avoiding reliance on explicit 3D inputs and reducing deployment costs. The code used in this paper is publicly available at: https://github.com/longkukuhi/RoboEye.

Zijun Long, Richard McCreadie, Muhammad Imran

In times of emergency, crisis response agencies need to quickly and accurately assess the situation on the ground in order to deploy relevant services and resources. However, authorities often have to make decisions based on limited information, as data on affected regions can be scarce until local response services can provide first-hand reports. Fortunately, the widespread availability of smartphones with high-quality cameras has made citizen journalism through social media a valuable source of information for crisis responders. However, analyzing the large volume of images posted by citizens requires more time and effort than is typically available. To address this issue, this paper proposes the use of state-of-the-art deep neural models for automatic image classification/tagging, specifically by adapting transformer-based architectures for crisis image classification (CrisisViT). We leverage the new Incidents1M crisis image dataset to develop a range of new transformer-based image classification models. Through experimentation over the standard Crisis image benchmark dataset, we demonstrate that the CrisisViT models significantly outperform previous approaches in emergency type, image relevance, humanitarian category, and damage severity classification. Additionally, we show that the new Incidents1M dataset can further augment the CrisisViT models resulting in an additional 1.25% absolute accuracy gain.

Zijun Long, George Killick, Lipeng Zhuang et al.

State-of-the-art pre-trained image models predominantly adopt a two-stage approach: initial unsupervised pre-training on large-scale datasets followed by task-specific fine-tuning using Cross-Entropy loss~(CE). However, it has been demonstrated that CE can compromise model generalization and stability. While recent works employing contrastive learning address some of these limitations by enhancing the quality of embeddings and producing better decision boundaries, they often overlook the importance of hard negative mining and rely on resource intensive and slow training using large sample batches. To counter these issues, we introduce a novel approach named CLCE, which integrates Label-Aware Contrastive Learning with CE. Our approach not only maintains the strengths of both loss functions but also leverages hard negative mining in a synergistic way to enhance performance. Experimental results demonstrate that CLCE significantly outperforms CE in Top-1 accuracy across twelve benchmarks, achieving gains of up to 3.52% in few-shot learning scenarios and 3.41% in transfer learning settings with the BEiT-3 model. Importantly, our proposed CLCE approach effectively mitigates the dependency of contrastive learning on large batch sizes such as 4096 samples per batch, a limitation that has previously constrained the application of contrastive learning in budget-limited hardware environments.

Zijun Long, Kangheng Liang, Gerardo Aragon-Camarasa et al.

Interactive Text-to-image retrieval (I-TIR) is an important enabler for a wide range of state-of-the-art services in domains such as e-commerce and education. However, current methods rely on finetuned Multimodal Large Language Models (MLLMs), which are costly to train and update, and exhibit poor generalizability. This latter issue is of particular concern, as: 1) finetuning narrows the pretrained distribution of MLLMs, thereby reducing generalizability; and 2) I-TIR introduces increasing query diversity and complexity. As a result, I-TIR solutions are highly likely to encounter queries and images not well represented in any training dataset. To address this, we propose leveraging Diffusion Models (DMs) for text-to-image mapping, to avoid finetuning MLLMs while preserving robust performance on complex queries. Specifically, we introduce Diffusion Augmented Retrieval (DAR), a framework that generates multiple intermediate representations via LLM-based dialogue refinements and DMs, producing a richer depiction of the user's information needs. This augmented representation facilitates more accurate identification of semantically and visually related images. Extensive experiments on four benchmarks show that for simple queries, DAR achieves results on par with finetuned I-TIR models, yet without incurring their tuning overhead. Moreover, as queries become more complex through additional conversational turns, DAR surpasses finetuned I-TIR models by up to 7.61% in Hits@10 after ten turns, illustrating its improved generalization for more intricate queries.

Zijun Long, Lipeng Zhuang, George Killick et al.

Human-annotated vision datasets inevitably contain a fraction of human mislabelled examples. While the detrimental effects of such mislabelling on supervised learning are well-researched, their influence on Supervised Contrastive Learning (SCL) remains largely unexplored. In this paper, we show that human-labelling errors not only differ significantly from synthetic label errors, but also pose unique challenges in SCL, different to those in traditional supervised learning methods. Specifically, our results indicate they adversely impact the learning process in the ~99% of cases when they occur as false positive samples. Existing noise-mitigating methods primarily focus on synthetic label errors and tackle the unrealistic setting of very high synthetic noise rates (40-80%), but they often underperform on common image datasets due to overfitting. To address this issue, we introduce a novel SCL objective with robustness to human-labelling errors, SCL-RHE. SCL-RHE is designed to mitigate the effects of real-world mislabelled examples, typically characterized by much lower noise rates (<5%). We demonstrate that SCL-RHE consistently outperforms state-of-the-art representation learning and noise-mitigating methods across various vision benchmarks, by offering improved resilience against human-labelling errors.

Zijun Long, George Killick, Richard McCreadie et al.

As Multimodal Large Language Models (MLLMs) grow in size, adapting them to specialized tasks becomes increasingly challenging due to high computational and memory demands. Indeed, traditional fine-tuning methods are costly, due to the need for extensive, task-specific training. While efficient adaptation methods exist that aim to reduce these costs, in practice they suffer from shallow inter-modal alignment, which severely hurts model effectiveness. To tackle these computational challenges and improve inter-modal alignment, we introduce the MultiWay-Adapter (MWA), a novel framework featuring an 'Alignment Enhancer'. This enhancer deepens inter-modal alignment, enabling high transferability with minimal tuning effort. Our experiments show that unlike prior efficient tuning approaches, MWA maintains model effectiveness, while reducing training time by up-to 57%. MWA is also lightweight, increasing model size by only 2-3% (in terms of parameters) for state-of-the-art foundation models like BEiT-3 Large. These results demonstrate that MWA provides an efficient and effective adaptation method for MLLMs, significantly broadening their applicability.