Pancreatic ductal adenocarcinoma (PDAC) is the third most common cause of cancer death in the United States. Predicting tumors like PDACs (including both classification and segmentation) from medical images by deep learning is becoming a growing trend, but usually a large number of annotated data are required for training, which is very labor-intensive and time-consuming. In this paper, we consider a partially supervised setting, where cheap image-level annotations are provided for all the training data, and the costly per-voxel annotations are only available for a subset of them. We propose an Inductive Attention Guidance Network (IAG-Net) to jointly learn a global image-level classifier for normal/PDAC classification and a local voxel-level classifier for semi-supervised PDAC segmentation. We instantiate both the global and the local classifiers by multiple instance learning (MIL), where the attention guidance, indicating roughly where the PDAC regions are, is the key to bridging them: For global MIL based normal/PDAC classification, attention serves as a weight for each instance (voxel) during MIL pooling, which eliminates the distraction from the background; For local MIL based semi-supervised PDAC segmentation, the attention guidance is inductive, which not only provides bag-level pseudo-labels to training data without per-voxel annotations for MIL training, but also acts as a proxy of an instance-level classifier. Experimental results show that our IAG-Net boosts PDAC segmentation accuracy by more than 5% compared with the state-of-the-arts.
도메인 일반화(Domain generalization)는 여러 도메인에 걸쳐 신뢰할 수 있는 분류를 지원하는 안정적인 표현을 식별해야 합니다. 도메인들은 기저 컴포넌트 간의 동일한 구조적 관계를 유지하면서도 색상, 질감 또는 시각적 스타일과 같은 저수준 속성에서는 차이가 날 수 있습니다. 기존 방법들은 주로 학습 과정을 개선하거나 도메인 간 특징을 정렬함으로써 이러한 차이를 해결합니다. 그러나 공유되는 구성적 구조를 암시적인 상태로 남겨두기 때문에, 더 신뢰할 수 있는 불변성의 원천을 간과할 수 있으며 결과적으로 보지 못한 도메인으로의 일반화 효과가 떨어질 수 있습니다. 본 논문에서는 시각적 인식을 시각적 프리미티브와 이들의 관계적 구성으로 분해하는 이미지 분류 프레임워크인 CRISP(Compositional Relational Invariance from Spatial Primitives)를 제안합니다. 우리는 이러한 구성을 프리미티브 위치 및 외형에 대한 소프트 단항, 이항, 삼항 술어로 표현하여, 엔드투엔드로 학습 가능한 공간적 및 시각적 정렬에 대한 미분 가능한 측정값을 산출합니다. 프리미티브와 관계적 구조를 공동으로 학습하기 위해, (1) 일반화된 특징을 추출하는 시각적 백본, (2) 이러한 특징을 미분 가능한 공간 좌표를 갖는 프리미티브 히트맵에 매핑하는 개념 병목 층(concept bottleneck layer), (3) 감지된 프리미티브 간의 후보 공간 관계를 평가하는 구조적 스코어링 층의 세 가지 구성 요소로 이루어진 엔드투엔드 아키텍처를 설계합니다. 마지막으로, 클래스별 관계 구성과 국소화된 프리미티브 외형의 결합 증거로부터 클래스 확률을 계산합니다. 널리 사용되는 DomainBed 스위트의 5가지 실제 이미지 분류 데이터셋에서 CRISP를 평가하며, 이는 묘사 스타일, 데이터셋 출처, 카메라 트랩 위치의 변화를 포괄하고 두 벤치마크 모두에서 새로운 최고 성능(state-of-the-art)을 달성합니다.
In medical imaging, there are clinically relevant segmentation tasks where the output mask is a projection to a subset of input image dimensions. In this work, we propose a novel convolutional neural network architecture that can effectively learn to produce a lower-dimensional segmentation mask than the input image. The network restores encoded representation only in a subset of input spatial dimensions and keeps the representation unchanged in the others. The newly proposed projective skip-connections allow linking the encoder and decoder in a UNet-like structure. We evaluated the proposed method on two clinically relevant tasks in retinal Optical Coherence Tomography (OCT): geographic atrophy and retinal blood vessel segmentation. The proposed method outperformed the current state-of-the-art approaches on all the OCT datasets used, consisting of 3D volumes and corresponding 2D en-face masks. The proposed architecture fills the methodological gap between image classification and ND image segmentation.
Artificial Neural Networks are connectionist systems that perform a given task by learning on examples without having prior knowledge about the task. This is done by finding an optimal point estimate for the weights in every node. Generally, the network using point estimates as weights perform well with large datasets, but they fail to express uncertainty in regions with little or no data, leading to overconfident decisions. In this paper, Bayesian Convolutional Neural Network (BayesCNN) using Variational Inference is proposed, that introduces probability distribution over the weights. Furthermore, the proposed BayesCNN architecture is applied to tasks like Image Classification, Image Super-Resolution and Generative Adversarial Networks. The results are compared to point-estimates based architectures on MNIST, CIFAR-10 and CIFAR-100 datasets for Image CLassification task, on BSD300 dataset for Image Super Resolution task and on CIFAR10 dataset again for Generative Adversarial Network task. BayesCNN is based on Bayes by Backprop which derives a variational approximation to the true posterior. We, therefore, introduce the idea of applying two convolutional operations, one for the mean and one for the variance. Our proposed method not only achieves performances equivalent to frequentist inference in identical architectures but also incorporate a measurement for uncertainties and regularisation. It further eliminates the use of dropout in the model. Moreover, we predict how certain the model prediction is based on the epistemic and aleatoric uncertainties and empirically show how the uncertainty can decrease, allowing the decisions made by the network to become more deterministic as the training accuracy increases. Finally, we propose ways to prune the Bayesian architecture and to make it more computational and time effective.
This study proposes a fully automated approach for the left atrial segmentation from routine cine long-axis cardiac magnetic resonance image sequences using deep convolutional neural networks and Bayesian filtering. The proposed approach consists of a classification network that automatically detects the type of long-axis sequence and three different convolutional neural network models followed by unscented Kalman filtering (UKF) that delineates the left atrium. Instead of training and predicting all long-axis sequence types together, the proposed approach first identifies the image sequence type as to 2, 3 and 4 chamber views, and then performs prediction based on neural nets trained for that particular sequence type. The datasets were acquired retrospectively and ground truth manual segmentation was provided by an expert radiologist. In addition to neural net based classification and segmentation, another neural net is trained and utilized to select image sequences for further processing using UKF to impose temporal consistency over cardiac cycle. A cyclic dynamic model with time-varying angular frequency is introduced in UKF to characterize the variations in cardiac motion during image scanning. The proposed approach was trained and evaluated separately with varying amount of training data with images acquired from 20, 40, 60 and 80 patients. Evaluations over 1515 images with equal number of images from each chamber group acquired from an additional 20 patients demonstrated that the proposed model outperformed state-of-the-art and yielded a mean Dice coefficient value of 94.1%, 93.7% and 90.1% for 2, 3 and 4-chamber sequences, respectively, when trained with datasets from 80 patients.
Vision-language models (VLMs) for radiology have emerged as a scalable paradigm by leveraging image-report pairs naturally produced in clinical workflows. However, this pairing reveals a mismatch in scale: each finding occupies only a small region of the image, yet supervision is provided only at the global image-report level. This poses a central challenge: prior approaches spread weight densely across all patches rather than concentrating on the sparse subset relevant to a given query. To address this, we present GLINT (Gated Language-Image alignmeNT), a framework that models this sparse correspondence. On the alignment side, we introduce Sparsely Gated Alignment, a novel architecture in which a sigmoid gate over a separate gate embedding space activates only the patches relevant to each textual query, yielding sparsity. On the representation side, we add Dense Feature Regularization, which anchors the trainable encoder's intermediate features to a frozen self-supervised learning (SSL) teacher, preserving the fine-grained patch features that the gate relies on. The same recipe applies to both 2D chest X-ray (CXR) and 3D chest computed tomography (CT), built with DINOv3 and V-JEPA 2.1, respectively. GLINT enables zero-shot classification, grounding, and segmentation from free-text queries, and to our knowledge is the first to demonstrate zero-shot segmentation on 3D CT volumes without mask supervision. Notably, the most pronounced gains arise on zero-shot grounding and segmentation, where sparse, query-specific localization is required, consistent with our design intent. In downstream evaluation, GLINT outperforms both SSL encoders and medical VLMs on classification, report generation, and segmentation. Code is available at https://github.com/deepnoid-ai/GLINT.
Most of the achievements in artificial intelligence so far were accomplished by supervised learning which requires numerous annotated training data and thus costs innumerable manpower for labeling. Unsupervised learning is one of the effective solutions to overcome such difficulties. In our work, we propose AugNet, a new deep learning training paradigm to learn image features from a collection of unlabeled pictures. We develop a method to construct the similarities between pictures as distance metrics in the embedding space by leveraging the inter-correlation between augmented versions of samples. Our experiments demonstrate that the method is able to represent the image in low dimensional space and performs competitively in downstream tasks such as image classification and image similarity comparison. Specifically, we achieved over 60% and 27% accuracy on the STL10 and CIFAR100 datasets with unsupervised clustering, respectively. Moreover, unlike many deep-learning-based image retrieval algorithms, our approach does not require access to external annotated datasets to train the feature extractor, but still shows comparable or even better feature representation ability and easy-to-use characteristics. In our evaluations, the method outperforms all the state-of-the-art image retrieval algorithms on some out-of-domain image datasets. The code for the model implementation is available at https://github.com/chenmingxiang110/AugNet.
Visual recognition is recently learned via either supervised learning on human-annotated image-label data or language-image contrastive learning with webly-crawled image-text pairs. While supervised learning may result in a more discriminative representation, language-image pretraining shows unprecedented zero-shot recognition capability, largely due to the different properties of data sources and learning objectives. In this work, we introduce a new formulation by combining the two data sources into a common image-text-label space. In this space, we propose a new learning paradigm, called Unified Contrastive Learning (UniCL) with a single learning objective to seamlessly prompt the synergy of two data types. Extensive experiments show that our UniCL is an effective way of learning semantically rich yet discriminative representations, universally for image recognition in zero-shot, linear-probe, fully finetuning and transfer learning scenarios. Particularly, it attains gains up to 9.2% and 14.5% in average on zero-shot recognition benchmarks over the language-image contrastive learning and supervised learning methods, respectively. In linear probe setting, it also boosts the performance over the two methods by 7.3% and 3.4%, respectively. Our study also indicates that UniCL stand-alone is a good learner on pure image-label data, rivaling the supervised learning methods across three image classification datasets and two types of vision backbones, ResNet and Swin Transformer. Code is available at https://github.com/microsoft/UniCL.
Filters of convolutional networks used in computer vision are often visualized as image patches that maximize the response of the filter. We use the same approach to interpret weight matrices in simple architectures for natural language processing tasks. We interpret a convolutional network for sentiment classification as word-based rules. Using the rule, we recover the performance of the original model.
Fine-grained image labels are desirable for many computer vision applications, such as visual search or mobile AI assistant. These applications rely on image classification models that can produce hundreds of thousands (e.g. 100K) of diversified fine-grained image labels on input images. However, training a network at this vocabulary scale is challenging, and suffers from intolerable large model size and slow training speed, which leads to unsatisfying classification performance. A straightforward solution would be training separate expert networks (specialists), with each specialist focusing on learning one specific vertical (e.g. cars, birds...). However, deploying dozens of expert networks in a practical system would significantly increase system complexity and inference latency, and consumes large amounts of computational resources. To address these challenges, we propose a Knowledge Concentration method, which effectively transfers the knowledge from dozens of specialists (multiple teacher networks) into one single model (one student network) to classify 100K object categories. There are three salient aspects in our method: (1) a multi-teacher single-student knowledge distillation framework; (2) a self-paced learning mechanism to allow the student to learn from different teachers at various paces; (3) structurally connected layers to expand the student network capacity with limited extra parameters. We validate our method on OpenImage and a newly collected dataset, Entity-Foto-Tree (EFT), with 100K categories, and show that the proposed model performs significantly better than the baseline generalist model.
Recent years have witnessed the great success of deep learning algorithms in the geoscience and remote sensing realm. Nevertheless, the security and robustness of deep learning models deserve special attention when addressing safety-critical remote sensing tasks. In this paper, we provide a systematic analysis of backdoor attacks for remote sensing data, where both scene classification and semantic segmentation tasks are considered. While most of the existing backdoor attack algorithms rely on visible triggers like squared patches with well-designed patterns, we propose a novel wavelet transform-based attack (WABA) method, which can achieve invisible attacks by injecting the trigger image into the poisoned image in the low-frequency domain. In this way, the high-frequency information in the trigger image can be filtered out in the attack, resulting in stealthy data poisoning. Despite its simplicity, the proposed method can significantly cheat the current state-of-the-art deep learning models with a high attack success rate. We further analyze how different trigger images and the hyper-parameters in the wavelet transform would influence the performance of the proposed method. Extensive experiments on four benchmark remote sensing datasets demonstrate the effectiveness of the proposed method for both scene classification and semantic segmentation tasks and thus highlight the importance of designing advanced backdoor defense algorithms to address this threat in remote sensing scenarios. The code will be available online at \url{https://github.com/ndraeger/waba}.
Models trained for classification often assume that all testing classes are known while training. As a result, when presented with an unknown class during testing, such closed-set assumption forces the model to classify it as one of the known classes. However, in a real world scenario, classification models are likely to encounter such examples. Hence, identifying those examples as unknown becomes critical to model performance. A potential solution to overcome this problem lies in a class of learning problems known as open-set recognition. It refers to the problem of identifying the unknown classes during testing, while maintaining performance on the known classes. In this paper, we propose an open-set recognition algorithm using class conditioned auto-encoders with novel training and testing methodology. In contrast to previous methods, training procedure is divided in two sub-tasks, 1. closed-set classification and, 2. open-set identification (i.e. identifying a class as known or unknown). Encoder learns the first task following the closed-set classification training pipeline, whereas decoder learns the second task by reconstructing conditioned on class identity. Furthermore, we model reconstruction errors using the Extreme Value Theory of statistical modeling to find the threshold for identifying known/unknown class samples. Experiments performed on multiple image classification datasets show proposed method performs significantly better than state of the art.
Classification based on Zero-shot Learning (ZSL) is the ability of a model to classify inputs into novel classes on which the model has not previously seen any training examples. Providing an auxiliary descriptor in the form of a set of attributes describing the new classes involved in the ZSL-based classification is one of the favored approaches to solving this challenging task. In this work, inspired by Hyperdimensional Computing (HDC), we propose the use of stationary binary codebooks of symbol-like distributed representations inside an attribute encoder to compactly represent a computationally simple end-to-end trainable model, which we name Hyperdimensional Computing Zero-shot Classifier~(HDC-ZSC). It consists of a trainable image encoder, an attribute encoder based on HDC, and a similarity kernel. We show that HDC-ZSC can be used to first perform zero-shot attribute extraction tasks and, can later be repurposed for Zero-shot Classification tasks with minimal architectural changes and minimal model retraining. HDC-ZSC achieves Pareto optimal results with a 63.8% top-1 classification accuracy on the CUB-200 dataset by having only 26.6 million trainable parameters. Compared to two other state-of-the-art non-generative approaches, HDC-ZSC achieves 4.3% and 9.9% better accuracy, while they require more than 1.85x and 1.72x parameters compared to HDC-ZSC, respectively.
Deep learning models designed for visual classification tasks on natural images have become prevalent in medical image analysis. However, medical images differ from typical natural images in many ways, such as significantly higher resolutions and smaller regions of interest. Moreover, both the global structure and local details play important roles in medical image analysis tasks. To address these unique properties of medical images, we propose a neural network that is able to classify breast cancer lesions utilizing information from both a global saliency map and multiple local patches. The proposed model outperforms the ResNet-based baseline and achieves radiologist-level performance in the interpretation of screening mammography. Although our model is trained only with image-level labels, it is able to generate pixel-level saliency maps that provide localization of possible malignant findings.
This paper investigates the role of global context for crowd counting. Specifically, a pure transformer is used to extract features with global information from overlapping image patches. Inspired by classification, we add a context token to the input sequence, to facilitate information exchange with tokens corresponding to image patches throughout transformer layers. Due to the fact that transformers do not explicitly model the tried-and-true channel-wise interactions, we propose a token-attention module (TAM) to recalibrate encoded features through channel-wise attention informed by the context token. Beyond that, it is adopted to predict the total person count of the image through regression-token module (RTM). Extensive experiments on various datasets, including ShanghaiTech, UCF-QNRF, JHU-CROWD++ and NWPU, demonstrate that the proposed context extraction techniques can significantly improve the performance over the baselines.
Pre-training on image-text colonoscopy records offers substantial potential for improving endoscopic image analysis, but faces challenges including non-informative background images, complex medical terminology, and ambiguous multi-lesion descriptions. We introduce Endo-CLIP, a novel self-supervised framework that enhances Contrastive Language-Image Pre-training (CLIP) for this domain. Endo-CLIP's three-stage framework--cleansing, attunement, and unification--addresses these challenges by (1) removing background frames, (2) leveraging large language models to extract clinical attributes for fine-grained contrastive learning, and (3) employing patient-level cross-attention to resolve multi-polyp ambiguities. Extensive experiments demonstrate that Endo-CLIP significantly outperforms state-of-the-art pre-training methods in zero-shot and few-shot polyp detection and classification, paving the way for more accurate and clinically relevant endoscopic analysis.
Sparse coding represents a signal sparsely by using an overcomplete dictionary, and obtains promising performance in practical computer vision applications, especially for signal restoration tasks such as image denoising and image inpainting. In recent years, many discriminative sparse coding algorithms have been developed for classification problems, but they cannot naturally handle visual data represented by multiview features. In addition, existing sparse coding algorithms use graph Laplacian to model the local geometry of the data distribution. It has been identified that Laplacian regularization biases the solution towards a constant function which possibly leads to poor extrapolating power. In this paper, we present multiview Hessian discriminative sparse coding (mHDSC) which seamlessly integrates Hessian regularization with discriminative sparse coding for multiview learning problems. In particular, mHDSC exploits Hessian regularization to steer the solution which varies smoothly along geodesics in the manifold, and treats the label information as an additional view of feature for incorporating the discriminative power for image annotation. We conduct extensive experiments on PASCAL VOC'07 dataset and demonstrate the effectiveness of mHDSC for image annotation.
The diagnosis of brain cancer relies heavily on medical imaging techniques, with MRI being the most commonly used. It is necessary to perform automatic segmentation of brain tumors on MRI images. This project intends to build an MRI algorithm based on U-Net. The residual network and the module used to enhance the context information are combined, and the void space convolution pooling pyramid is added to the network for processing. The brain glioma MRI image dataset provided by cancer imaging archives was experimentally verified. A multi-scale segmentation method based on a weighted least squares filter was used to complete the 3D reconstruction of brain tumors. Thus, the accuracy of three-dimensional reconstruction is further improved. Experiments show that the local texture features obtained by the proposed algorithm are similar to those obtained by laser scanning. The algorithm is improved by using the U-Net method and an accuracy of 0.9851 is obtained. This approach significantly enhances the precision of image segmentation and boosts the efficiency of image classification.
Generative AI is transforming image synthesis, enabling the creation of high-quality, diverse, and photorealistic visuals across industries like design, media, healthcare, and autonomous systems. Advances in techniques such as image-to-image translation, text-to-image generation, domain transfer, and multimodal alignment have broadened the scope of automated visual content creation, supporting a wide spectrum of applications. These advancements are driven by models like Generative Adversarial Networks (GANs), conditional frameworks, and diffusion-based approaches such as Stable Diffusion. This work presents a structured classification of image generation techniques based on the nature of the input, organizing methods by input modalities like noisy vectors, latent representations, and conditional inputs. We explore the principles behind these models, highlight key frameworks including DALL-E, ControlNet, and DeepSeek Janus-Pro, and address challenges such as computational costs, data biases, and output alignment with user intent. By offering this input-centric perspective, this study bridges technical depth with practical insights, providing researchers and practitioners with a comprehensive resource to harness generative AI for real-world applications.
We introduce a novel principle for self-supervised feature learning based on the discrimination of specific transformations of an image. We argue that the generalization capability of learned features depends on what image neighborhood size is sufficient to discriminate different image transformations: The larger the required neighborhood size and the more global the image statistics that the feature can describe. An accurate description of global image statistics allows to better represent the shape and configuration of objects and their context, which ultimately generalizes better to new tasks such as object classification and detection. This suggests a criterion to choose and design image transformations. Based on this criterion, we introduce a novel image transformation that we call limited context inpainting (LCI). This transformation inpaints an image patch conditioned only on a small rectangular pixel boundary (the limited context). Because of the limited boundary information, the inpainter can learn to match local pixel statistics, but is unlikely to match the global statistics of the image. We claim that the same principle can be used to justify the performance of transformations such as image rotations and warping. Indeed, we demonstrate experimentally that learning to discriminate transformations such as LCI, image warping and rotations, yields features with state of the art generalization capabilities on several datasets such as Pascal VOC, STL-10, CelebA, and ImageNet. Remarkably, our trained features achieve a performance on Places on par with features trained through supervised learning with ImageNet labels.