Visual Place Recognition (VPR) aims to robustly identify locations by leveraging image retrieval based on descriptors encoded from environmental images. However, drastic appearance changes of images captured from different viewpoints at the same location pose incoherent supervision signals for descriptor learning, which severely hinder the performance of VPR. Previous work proposes classifying images based on manually defined rules or ground truth labels for viewpoints, followed by descriptor training based on the classification results. However, not all datasets have ground truth labels of viewpoints and manually defined rules may be suboptimal, leading to degraded descriptor performance.To address these challenges, we introduce the mutual learning of viewpoint self-classification and VPR. Starting from coarse classification based on geographical coordinates, we progress to finer classification of viewpoints using simple clustering techniques. The dataset is partitioned in an unsupervised manner while simultaneously training a descriptor extractor for place recognition. Experimental results show that this approach almost perfectly partitions the dataset based on viewpoints, thus achieving mutually reinforcing effects. Our method even excels state-of-the-art (SOTA) methods that partition datasets using ground truth labels.
Deep supervised hashing has become a pivotal technique in large-scale image retrieval, offering significant benefits in terms of storage and search efficiency. However, existing deep supervised hashing models predominantly focus on generating fixed-length hash codes. This approach fails to address the inherent trade-off between efficiency and effectiveness when using hash codes of varying lengths. To determine the optimal hash code length for a specific task, multiple models must be trained for different lengths, leading to increased training time and computational overhead. Furthermore, the current paradigm overlooks the potential relationships between hash codes of different lengths, limiting the overall effectiveness of the models. To address these challenges, we propose the Nested Hash Layer (NHL), a plug-and-play module designed for existing deep supervised hashing models. The NHL framework introduces a novel mechanism to simultaneously generate hash codes of varying lengths in a nested manner. To tackle the optimization conflicts arising from the multiple learning objectives associated with different code lengths, we further propose an adaptive weights strategy that dynamically monitors and adjusts gradients during training. Additionally, recognizing that the structural information in longer hash codes can provide valuable guidance for shorter hash codes, we develop a long-short cascade self-distillation method within the NHL to enhance the overall quality of the generated hash codes. Extensive experiments demonstrate that NHL not only accelerates the training process but also achieves superior retrieval performance across various deep hashing models. Our code is publicly available at https://github.com/hly1998/NHL.
Image retrieval methods rely on metric learning to train backbone feature extraction models that can extract discriminant queries and reference (gallery) feature representations for similarity matching. Although state-of-the-art accuracy has improved considerably with the advent of deep learning (DL) models trained on large datasets, image retrieval remains challenging in many real-world video analytics and surveillance applications, e.g., person re-identification. Using the Euclidean space for matching limits the performance in real-world applications due to the curse of dimensionality, overfitting, and sensitivity to noisy data. We argue that the feature dissimilarity space is more suitable for similarity matching, and propose a dichotomy transformation to project query and reference embeddings into a single embedding in the dissimilarity space. We also advocate for end-to-end training of a backbone and binary classification models for pair-wise matching. As opposed to comparing the distance between queries and reference embeddings, we show the benefits of classifying the single dissimilarity space embedding (as similar or dissimilar), especially when trained end-to-end. We propose a method to train the max-margin classifier together with the backbone feature extractor by applying constraints to the L2 norm of the classifier weights along with the hinge loss. Our extensive experiments on challenging image retrieval datasets and using diverse feature extraction backbones highlight the benefits of similarity matching in the dissimilarity space. In particular, when jointly training the feature extraction backbone and regularised classifier for matching, the dissimilarity space provides a higher level of accuracy.
Purpose: Intraoperative ultrasound (US) can enhance real-time visualization in transoral robotic surgery. The surgeon creates a mental map with a pre-operative scan. Then, a surgical assistant performs freehand US scanning during the surgery while the surgeon operates at the remote surgical console. Communicating the target scanning plane in the surgeon's mental map is difficult. Automatic image retrieval can help match intraoperative images to preoperative scans, guiding the assistant to adjust the US probe toward the target plane. Methods: We propose a self-supervised contrastive learning approach to match intraoperative US views to a preoperative image database. We introduce a novel contrastive learning strategy that leverages intra-sweep similarity and US probe location to improve feature encoding. Additionally, our model incorporates a flexible threshold to reject unsatisfactory matches. Results: Our method achieves 92.30% retrieval accuracy on simulated data and outperforms state-of-the-art temporal-based contrastive learning approaches. Our ablation study demonstrates that using probe location in the optimization goal improves image representation, suggesting that semantic information can be extracted from probe location. We also present our approach on real patient data to show the feasibility of the proposed US probe localization system despite tissue deformation from tongue retraction. Conclusion: Our contrastive learning method, which utilizes intra-sweep similarity and US probe location, enhances US image representation learning. We also demonstrate the feasibility of using our image retrieval method to provide neck US localization on real patient US after tongue retraction.
Composed Image Retrieval (CIR) involves retrieving a target image based on a composed query of an image paired with text that specifies modifications or changes to the visual reference. CIR is inherently an instruction-following task, as the model needs to interpret and apply modifications to the image. In practice, due to the scarcity of annotated data in downstream tasks, Zero-Shot CIR (ZS-CIR) is desirable. While existing ZS-CIR models based on CLIP have shown promising results, their capability in interpreting and following modification instructions remains limited. Some research attempts to address this by incorporating Large Language Models (LLMs). However, these approaches still face challenges in effectively integrating multimodal information and instruction understanding. To tackle above challenges, we propose a novel embedding method utilizing an instruction-tuned Multimodal LLM (MLLM) to generate composed representation, which significantly enhance the instruction following capability for a comprehensive integration between images and instructions. Nevertheless, directly applying MLLMs introduces a new challenge since MLLMs are primarily designed for text generation rather than embedding extraction as required in CIR. To address this, we introduce a two-stage training strategy to efficiently learn a joint multimodal embedding space and further refining the ability to follow modification instructions by tuning the model in a triplet dataset similar to the CIR format. Extensive experiments on four public datasets: FashionIQ, CIRR, GeneCIS, and CIRCO demonstrates the superior performance of our model, outperforming state-of-the-art baselines by a significant margin. Codes are available at the GitHub repository.
Medical image understanding requires meticulous examination of fine visual details, with particular regions requiring additional attention. While radiologists build such expertise over years of experience, it is challenging for AI models to learn where to look with limited amounts of training data. This limitation results in unsatisfying robustness in medical image understanding. To address this issue, we propose Diffusion-based Feature Augmentation (DAug), a portable method that improves a perception model's performance with a generative model's output. Specifically, we extend a radiology image to multiple channels, with the additional channels being the heatmaps of regions where diseases tend to develop. A diffusion-based image-to-image translation model was used to generate such heatmaps conditioned on selected disease classes. Our method is motivated by the fact that generative models learn the distribution of normal and abnormal images, and such knowledge is complementary to image understanding tasks. In addition, we propose the Image-Text-Class Hybrid Contrastive learning to utilize both text and class labels. With two novel approaches combined, our method surpasses baseline models without changing the model architecture, and achieves state-of-the-art performance on both medical image retrieval and classification tasks.
This work addresses composed image retrieval in the context of domain conversion, where the content of a query image is retrieved in the domain specified by the query text. We show that a strong vision-language model provides sufficient descriptive power without additional training. The query image is mapped to the text input space using textual inversion. Unlike common practice that invert in the continuous space of text tokens, we use the discrete word space via a nearest-neighbor search in a text vocabulary. With this inversion, the image is softly mapped across the vocabulary and is made more robust using retrieval-based augmentation. Database images are retrieved by a weighted ensemble of text queries combining mapped words with the domain text. Our method outperforms prior art by a large margin on standard and newly introduced benchmarks. Code: https://github.com/NikosEfth/freedom
Multi-modal datasets, like those involving images, often miss the detailed descriptions that properly capture the rich information encoded in each item. This makes answering complex natural language queries a major challenge in this domain. In particular, unlike the traditional nearest neighbor search, where the tuples and the query are represented as points in a single metric space, these settings involve queries and tuples embedded in fundamentally different spaces, making the traditional query answering methods inapplicable. Existing literature addresses this challenge for image datasets through vector representations jointly trained on natural language and images. This technique, however, underperforms for complex queries due to various reasons. This paper takes a step towards addressing this challenge by introducing a Generative-based Monte Carlo method that utilizes foundation models to generate synthetic samples that capture the complexity of the natural language query and represent it in the same metric space as the multi-modal data. Following this method, we propose Needle, a database for image data retrieval. Instead of relying on contrastive learning or metadata-searching approaches, our system is based on synthetic data generation to capture the complexities of natural language queries. Our system is open-source and ready for deployment, designed to be easily adopted by researchers and developers. The comprehensive experiments on various benchmark datasets verify that this system significantly outperforms state-of-the-art text-to-image retrieval methods in the literature. Any foundation model and embedder can be easily integrated into Needle to improve the performance, piggybacking on the advancements in these technologies.
Approximate Nearest Neighbor search is one of the keys to high-scale data retrieval performance in many applications. The work is a bridge between feature extraction and ANN indexing through fine-tuning a ResNet50 model with various ANN methods: FAISS and Annoy. We evaluate the systems with respect to indexing time, memory usage, query time, precision, recall, F1-score, and Recall@5 on a custom image dataset. FAISS's Product Quantization can achieve a precision of 98.40% with low memory usage at 0.24 MB index size, and Annoy is the fastest, with average query times of 0.00015 seconds, at a slight cost to accuracy. These results reveal trade-offs among speed, accuracy, and memory efficiency and offer actionable insights into the optimization of feature-based image retrieval systems. This study will serve as a blueprint for constructing actual retrieval pipelines and be built on fine-tuned deep learning networks and associated ANN methods.
Active Learning (AL) is a user-interactive approach aimed at reducing annotation costs by selecting the most crucial examples to label. Although AL has been extensively studied for image classification tasks, the specific scenario of interactive image retrieval has received relatively little attention. This scenario presents unique characteristics, including an open-set and class-imbalanced binary classification, starting with very few labeled samples. We introduce a novel batch-mode Active Learning framework named GAL (Greedy Active Learning) that better copes with this application. It incorporates a new acquisition function for sample selection that measures the impact of each unlabeled sample on the classifier. We further embed this strategy in a greedy selection approach, better exploiting the samples within each batch. We evaluate our framework with both linear (SVM) and non-linear MLP/Gaussian Process classifiers. For the Gaussian Process case, we show a theoretical guarantee on the greedy approximation. Finally, we assess our performance for the interactive content-based image retrieval task on several benchmarks and demonstrate its superiority over existing approaches and common baselines. Code is available at https://github.com/barleah/GreedyAL.
In this paper, we present a Neuron Abandoning Attention Flow (NAFlow) method to address the open problem of visually explaining the attention evolution dynamics inside CNNs when making their classification decisions. A novel cascading neuron abandoning back-propagation algorithm is designed to trace neurons in all layers of a CNN that involve in making its prediction to address the problem of significant interference from abandoned neurons. Firstly, a Neuron Abandoning Back-Propagation (NA-BP) module is proposed to generate Back-Propagated Feature Maps (BPFM) by using the inverse function of the intermediate layers of CNN models, on which the neurons not used for decision-making are abandoned. Meanwhile, the cascading NA-BP modules calculate the tensors of importance coefficients which are linearly combined with the tensors of BPFMs to form the NAFlow. Secondly, to be able to visualize attention flow for similarity metric-based CNN models, a new channel contribution weights module is proposed to calculate the importance coefficients via Jacobian Matrix. The effectiveness of the proposed NAFlow is validated on nine widely-used CNN models for various tasks of general image classification, contrastive learning classification, few-shot image classification, and image retrieval.
The Zero-shot Composed Image Retrieval (ZSCIR) requires retrieving images that match the query image and the relative captions. Current methods focus on projecting the query image into the text feature space, subsequently combining them with features of query texts for retrieval. However, retrieving images only with the text features cannot guarantee detailed alignment due to the natural gap between images and text. In this paper, we introduce Imagined Proxy for CIR (IP-CIR), a training-free method that creates a proxy image aligned with the query image and text description, enhancing query representation in the retrieval process. We first leverage the large language model's generalization capability to generate an image layout, and then apply both the query text and image for conditional generation. The robust query features are enhanced by merging the proxy image, query image, and text semantic perturbation. Our newly proposed balancing metric integrates text-based and proxy retrieval similarities, allowing for more accurate retrieval of the target image while incorporating image-side information into the process. Experiments on three public datasets demonstrate that our method significantly improves retrieval performances. We achieve state-of-the-art (SOTA) results on the CIRR dataset with a Recall@K of 70.07 at K=10. Additionally, we achieved an improvement in Recall@10 on the FashionIQ dataset, rising from 45.11 to 45.74, and improved the baseline performance in CIRCO with a mAPK@10 score, increasing from 32.24 to 34.26.
Diffusion models have recently been employed to generate high-quality images, reducing the need for manual data collection and improving model generalization in tasks such as object detection, instance segmentation, and image perception. However, the synthetic framework is usually designed with meticulous human effort for each task due to various requirements on image layout, content, and annotation formats, restricting the application of synthetic data on more general scenarios. In this paper, we propose AnySynth, a unified framework integrating adaptable, comprehensive, and highly controllable components capable of generating an arbitrary type of synthetic data given diverse requirements. Specifically, the Task-Specific Layout Generation Module is first introduced to produce reasonable layouts for different tasks by leveraging the generation ability of large language models and layout priors of real-world images. A Uni-Controlled Image Generation Module is then developed to create high-quality synthetic images that are controllable and based on the generated layouts. In addition, user specific reference images, and style images can be incorporated into the generation to task requirements. Finally, the Task-Oriented Annotation Module offers precise and detailed annotations for the generated images across different tasks. We have validated our framework's performance across various tasks, including Few-shot Object Detection, Cross-domain Object Detection, Zero-shot Composed Image Retrieval, and Multi-modal Image Perception and Grounding. The specific data synthesized by our framework significantly improves model performance in these tasks, demonstrating the generality and effectiveness of our framework.
Structuring latent representations in a hierarchical manner enables models to learn patterns at multiple levels of abstraction. However, most prevalent image understanding models focus on visual similarity, and learning visual hierarchies is relatively unexplored. In this work, for the first time, we introduce a learning paradigm that can encode user-defined multi-level visual hierarchies in hyperbolic space without requiring explicit hierarchical labels. As a concrete example, first, we define a part-based image hierarchy using object-level annotations within and across images. Then, we introduce an approach to enforce the hierarchy using contrastive loss with pairwise entailment metrics. Finally, we discuss new evaluation metrics to effectively measure hierarchical image retrieval. Encoding these complex relationships ensures that the learned representations capture semantic and structural information that transcends mere visual similarity. Experiments in part-based image retrieval show significant improvements in hierarchical retrieval tasks, demonstrating the capability of our model in capturing visual hierarchies.
Generative methods now produce outputs nearly indistinguishable from real data but often fail to fully capture the data distribution. Unlike quality issues, diversity limitations in generative models are hard to detect visually, requiring specific metrics for assessment. In this paper, we draw attention to the current lack of diversity in generative models and the inability of common metrics to measure this. We achieve this by framing diversity as an image retrieval problem, where we measure how many real images can be retrieved using synthetic data as queries. This yields the Image Retrieval Score (IRS), an interpretable, hyperparameter-free metric that quantifies the diversity of a generative model's output. IRS requires only a subset of synthetic samples and provides a statistical measure of confidence. Our experiments indicate that current feature extractors commonly used in generative model assessment are inadequate for evaluating diversity effectively. Consequently, we perform an extensive search for the best feature extractors to assess diversity. Evaluation reveals that current diffusion models converge to limited subsets of the real distribution, with no current state-of-the-art models superpassing 77% of the diversity of the training data. To address this limitation, we introduce Diversity-Aware Diffusion Models (DiADM), a novel approach that improves diversity of unconditional diffusion models without loss of image quality. We do this by disentangling diversity from image quality by using a diversity aware module that uses pseudo-unconditional features as input. We provide a Python package offering unified feature extraction and metric computation to further facilitate the evaluation of generative models https://github.com/MischaD/beyondfid.
Cross-view Geo-localisation is typically performed at a coarse granularity, because densely sampled satellite image patches overlap heavily. This heavy overlap would make disambiguating patches very challenging. However, by opting for sparsely sampled patches, prior work has placed an artificial upper bound on the localisation accuracy that is possible. Even a perfect oracle system cannot achieve accuracy greater than the average separation of the tiles. To solve this limitation, we propose combining cross-view geo-localisation and relative pose estimation to increase precision to a level practical for real-world application. We develop PEnG, a 2-stage system which first predicts the most likely edges from a city-scale graph representation upon which a query image lies. It then performs relative pose estimation within these edges to determine a precise position. PEnG presents the first technique to utilise both viewpoints available within cross-view geo-localisation datasets to enhance precision to a sub-metre level, with some examples achieving centimetre level accuracy. Our proposed ensemble achieves state-of-the-art precision - with relative Top-5m retrieval improvements on previous works of 213%. Decreasing the median euclidean distance error by 96.90% from the previous best of 734m down to 22.77m, when evaluating with 90 degree horizontal FOV images. Code will be made available: tavisshore.co.uk/PEnG
Remote sensing cross-modal text-image retrieval (RSCTIR) has gained attention for its utility in information mining. However, challenges remain in effectively integrating global and local information due to variations in remote sensing imagery and ensuring proper feature pre-alignment before modal fusion, which affects retrieval accuracy and efficiency. To address these issues, we propose CMPAGL, a cross-modal pre-aligned method leveraging global and local information. Our Gswin transformer block combines local window self-attention and global-local window cross-attention to capture multi-scale features. A pre-alignment mechanism simplifies modal fusion training, improving retrieval performance. Additionally, we introduce a similarity matrix reweighting (SMR) algorithm for reranking, and enhance the triplet loss function with an intra-class distance term to optimize feature learning. Experiments on four datasets, including RSICD and RSITMD, validate CMPAGL's effectiveness, achieving up to 4.65% improvement in R@1 and 2.28% in mean Recall (mR) over state-of-the-art methods.
Compositional reasoning is a hallmark of human visual intelligence. Yet, despite the size of large vision-language models, they struggle to represent simple compositions by combining objects with their attributes. To measure this lack of compositional capability, we design Cola, a text-to-image retrieval benchmark to Compose Objects Localized with Attributes. To solve Cola, a model must retrieve images with the correct configuration of attributes and objects and avoid choosing a distractor image with the same objects and attributes but in the wrong configuration. Cola contains about 1.2k composed queries of 168 objects and 197 attributes on around 30K images. Our human evaluation finds that Cola is 83.33% accurate, similar to contemporary compositionality benchmarks. Using Cola as a testbed, we explore empirical modeling designs to adapt pre-trained vision-language models to reason compositionally. We explore 6 adaptation strategies on 2 seminal vision-language models, using compositionality-centric test benchmarks - Cola and CREPE. We find the optimal adaptation strategy is to train a multi-modal attention layer that jointly attends over the frozen pre-trained image and language features. Surprisingly, training multimodal layers on CLIP performs better than tuning a larger FLAVA model with already pre-trained multimodal layers. Furthermore, our adaptation strategy improves CLIP and FLAVA to comparable levels, suggesting that training multimodal layers using contrastive attribute-object data is key, as opposed to using them pre-trained. Lastly, we show that Cola is harder than a closely related contemporary benchmark, CREPE, since simpler fine-tuning strategies without multimodal layers suffice on CREPE but not on Cola. However, we still see a significant gap between our best adaptation and human accuracy, suggesting considerable room for further research.
Hard negative generation aims to generate informative negative samples that help to determine the decision boundaries and thus facilitate advancing deep metric learning. Current works select pair/triplet samples, learn their correlations, and fuse them to generate hard negatives. However, these works merely consider the local correlations of selected samples, ignoring global sample correlations that would provide more significant information to generate more informative negatives. In this work, we propose a Globally Correlation-Aware Hard Negative Generation (GCA-HNG) framework, which first learns sample correlations from a global perspective and exploits these correlations to guide generating hardness-adaptive and diverse negatives. Specifically, this approach begins by constructing a structured graph to model sample correlations, where each node represents a specific sample and each edge represents the correlations between corresponding samples. Then, we introduce an iterative graph message propagation to propagate the messages of node and edge through the whole graph and thus learn the sample correlations globally. Finally, with the guidance of the learned global correlations, we propose a channel-adaptive manner to combine an anchor and multiple negatives for HNG. Compared to current methods, GCA-HNG allows perceiving sample correlations with numerous negatives from a global and comprehensive perspective and generates the negatives with better hardness and diversity. Extensive experiment results demonstrate that the proposed GCA-HNG is superior to related methods on four image retrieval benchmark datasets. Codes and trained models are available at \url{https://github.com/PWenJay/GCA-HNG}.
Zero-Shot Composed Image Retrieval (ZS-CIR) has attracted more attention in recent years, focusing on retrieving a specific image based on a query composed of a reference image and a relative text without training samples. Specifically, the relative text describes the differences between the two images. Prevailing ZS-CIR methods employ image-to-text (I2T) models to convert the query image into a single caption, which is further merged with the relative text by text-fusion approaches to form a composed text for retrieval. However, these methods neglect the fact that ZS-CIR entails considering not only the final similarity between the composed text and retrieved images but also the semantic increment during the compositional editing process. To address this limitation, this paper proposes a training-free method called Semantic Editing Increment for ZS-CIR (SEIZE) to retrieve the target image based on the query image and text without training. Firstly, we employ a pre-trained captioning model to generate diverse captions for the reference image and prompt Large Language Models (LLMs) to perform breadth compositional reasoning based on these captions and relative text, thereby covering the potential semantics of the target image. Then, we design a semantic editing search to incorporate the semantic editing increment contributed by the relative text into the retrieval process. Concretely, we comprehensively consider relative semantic increment and absolute similarity as the final retrieval score, which is subsequently utilized to retrieve the target image in the CLIP feature space. Extensive experiments on three public datasets demonstrate that our proposed SEIZE achieves the new state-of-the-art performance. The code is publicly available at https://github.com/yzy-bupt/SEIZE.