Vision-language models (VLMs) seamlessly integrate visual and textual data to perform tasks such as image classification, caption generation, and visual question answering. However, adversarial images often struggle to deceive all prompts effectively in the context of cross-prompt migration attacks, as the probability distribution of the tokens in these images tends to favor the semantics of the original image rather than the target tokens. To address this challenge, we propose a Contextual-Injection Attack (CIA) that employs gradient-based perturbation to inject target tokens into both visual and textual contexts, thereby improving the probability distribution of the target tokens. By shifting the contextual semantics towards the target tokens instead of the original image semantics, CIA enhances the cross-prompt transferability of adversarial images.Extensive experiments on the BLIP2, InstructBLIP, and LLaVA models show that CIA outperforms existing methods in cross-prompt transferability, demonstrating its potential for more effective adversarial strategies in VLMs.
We propose CLIP-EBC, the first fully CLIP-based model for accurate crowd density estimation. While the CLIP model has demonstrated remarkable success in addressing recognition tasks such as zero-shot image classification, its potential for counting has been largely unexplored due to the inherent challenges in transforming a regression problem, such as counting, into a recognition task. In this work, we investigate and enhance CLIP's ability to count, focusing specifically on the task of estimating crowd sizes from images. Existing classification-based crowd-counting frameworks have significant limitations, including the quantization of count values into bordering real-valued bins and the sole focus on classification errors. These practices result in label ambiguity near the shared borders and inaccurate prediction of count values. Hence, directly applying CLIP within these frameworks may yield suboptimal performance. To address these challenges, we first propose the Enhanced Blockwise Classification (EBC) framework. Unlike previous methods, EBC utilizes integer-valued bins, effectively reducing ambiguity near bin boundaries. Additionally, it incorporates a regression loss based on density maps to improve the prediction of count values. Within our backbone-agnostic EBC framework, we then introduce CLIP-EBC to fully leverage CLIP's recognition capabilities for this task. Extensive experiments demonstrate the effectiveness of EBC and the competitive performance of CLIP-EBC. Specifically, our EBC framework can improve existing classification-based methods by up to 44.5% on the UCF-QNRF dataset, and CLIP-EBC achieves state-of-the-art performance on the NWPU-Crowd test set, with an MAE of 58.2 and an RMSE of 268.5, representing improvements of 8.6% and 13.3% over the previous best method, STEERER. The code and weights are available at https://github.com/Yiming-M/CLIP-EBC.
Convolutional neural networks (CNNs) are similar to "ordinary" neural networks in the sense that they are made up of hidden layers consisting of neurons with "learnable" parameters. These neurons receive inputs, performs a dot product, and then follows it with a non-linearity. The whole network expresses the mapping between raw image pixels and their class scores. Conventionally, the Softmax function is the classifier used at the last layer of this network. However, there have been studies (Alalshekmubarak and Smith, 2013; Agarap, 2017; Tang, 2013) conducted to challenge this norm. The cited studies introduce the usage of linear support vector machine (SVM) in an artificial neural network architecture. This project is yet another take on the subject, and is inspired by (Tang, 2013). Empirical data has shown that the CNN-SVM model was able to achieve a test accuracy of ~99.04% using the MNIST dataset (LeCun, Cortes, and Burges, 2010). On the other hand, the CNN-Softmax was able to achieve a test accuracy of ~99.23% using the same dataset. Both models were also tested on the recently-published Fashion-MNIST dataset (Xiao, Rasul, and Vollgraf, 2017), which is suppose to be a more difficult image classification dataset than MNIST (Zalandoresearch, 2017). This proved to be the case as CNN-SVM reached a test accuracy of ~90.72%, while the CNN-Softmax reached a test accuracy of ~91.86%. The said results may be improved if data preprocessing techniques were employed on the datasets, and if the base CNN model was a relatively more sophisticated than the one used in this study.
Deep learning could be prone to learning shortcuts raised by dataset bias and result in inaccurate, unreliable, and unfair models, which impedes its adoption in real-world clinical applications. Despite its significance, there is a dearth of research in the medical image classification domain to address dataset bias. Furthermore, the bias labels are often agnostic, as identifying biases can be laborious and depend on post-hoc interpretation. This paper proposes learning Adaptive Agreement from a Biased Council (Ada-ABC), a debiasing framework that does not rely on explicit bias labels to tackle dataset bias in medical images. Ada-ABC develops a biased council consisting of multiple classifiers optimized with generalized cross entropy loss to learn the dataset bias. A debiasing model is then simultaneously trained under the guidance of the biased council. Specifically, the debiasing model is required to learn adaptive agreement with the biased council by agreeing on the correctly predicted samples and disagreeing on the wrongly predicted samples by the biased council. In this way, the debiasing model could learn the target attribute on the samples without spurious correlations while also avoiding ignoring the rich information in samples with spurious correlations. We theoretically demonstrated that the debiasing model could learn the target features when the biased model successfully captures dataset bias. Moreover, to our best knowledge, we constructed the first medical debiasing benchmark from four datasets containing seven different bias scenarios. Our extensive experiments practically showed that our proposed Ada-ABC outperformed competitive approaches, verifying its effectiveness in mitigating dataset bias for medical image classification. The codes and organized benchmark datasets will be made publicly available.
The success of deep learning is inseparable from normalization layers. Researchers have proposed various normalization functions, and each of them has both advantages and disadvantages. In response, efforts have been made to design a unified normalization function that combines all normalization procedures and mitigates their weaknesses. We also proposed a new normalization function called Adaptive Fusion Normalization. Through experiments, we demonstrate AFN outperforms the previous normalization techniques in domain generalization and image classification tasks.
Deep convolutional neural networks (CNN) have revolutionized various fields of vision research and have seen unprecedented adoption for multiple tasks such as classification, detection, captioning, etc. However, they offer little transparency into their inner workings and are often treated as black boxes that deliver excellent performance. In this work, we aim at alleviating this opaqueness of CNNs by providing visual explanations for the network's predictions. Our approach can analyze variety of CNN based models trained for vision applications such as object recognition and caption generation. Unlike existing methods, we achieve this via unraveling the forward pass operation. Proposed method exploits feature dependencies across the layer hierarchy and uncovers the discriminative image locations that guide the network's predictions. We name these locations CNN-Fixations, loosely analogous to human eye fixations. Our approach is a generic method that requires no architectural changes, additional training or gradient computation and computes the important image locations (CNN Fixations). We demonstrate through a variety of applications that our approach is able to localize the discriminative image locations across different network architectures, diverse vision tasks and data modalities.
The exponential growth of large language models (LLMs) has opened up numerous possibilities for multimodal AGI systems. However, the progress in vision and vision-language foundation models, which are also critical elements of multi-modal AGI, has not kept pace with LLMs. In this work, we design a large-scale vision-language foundation model (InternVL), which scales up the vision foundation model to 6 billion parameters and progressively aligns it with the LLM, using web-scale image-text data from various sources. This model can be broadly applied to and achieve state-of-the-art performance on 32 generic visual-linguistic benchmarks including visual perception tasks such as image-level or pixel-level recognition, vision-language tasks such as zero-shot image/video classification, zero-shot image/video-text retrieval, and link with LLMs to create multi-modal dialogue systems. It has powerful visual capabilities and can be a good alternative to the ViT-22B. We hope that our research could contribute to the development of multi-modal large models. Code and models are available at https://github.com/OpenGVLab/InternVL.
This paper addresses the few-shot image classification problem, where the classification task is performed on unlabeled query samples given a small amount of labeled support samples only. One major challenge of the few-shot learning problem is the large variety of object visual appearances that prevents the support samples to represent that object comprehensively. This might result in a significant difference between support and query samples, therefore undermining the performance of few-shot algorithms. In this paper, we tackle the problem by proposing Few-shot Cosine Transformer (FS-CT), where the relational map between supports and queries is effectively obtained for the few-shot tasks. The FS-CT consists of two parts, a learnable prototypical embedding network to obtain categorical representations from support samples with hard cases, and a transformer encoder to effectively achieve the relational map from two different support and query samples. We introduce Cosine Attention, a more robust and stable attention module that enhances the transformer module significantly and therefore improves FS-CT performance from 5% to over 20% in accuracy compared to the default scaled dot-product mechanism. Our method performs competitive results in mini-ImageNet, CUB-200, and CIFAR-FS on 1-shot learning and 5-shot learning tasks across backbones and few-shot configurations. We also developed a custom few-shot dataset for Yoga pose recognition to demonstrate the potential of our algorithm for practical application. Our FS-CT with cosine attention is a lightweight, simple few-shot algorithm that can be applied for a wide range of applications, such as healthcare, medical, and security surveillance. The official implementation code of our Few-shot Cosine Transformer is available at https://github.com/vinuni-vishc/Few-Shot-Cosine-Transformer
Image classification problems are typically addressed by first collecting examples with candidate labels, second cleaning the candidate labels manually, and third training a deep neural network on the clean examples. The manual labeling step is often the most expensive one as it requires workers to label millions of images. In this paper we propose to work without any explicitly labeled data by i) directly training the deep neural network on the noisy candidate labels, and ii) early stopping the training to avoid overfitting. With this procedure we exploit an intriguing property of standard overparameterized convolutional neural networks trained with (stochastic) gradient descent: Clean labels are fitted faster than noisy ones. We consider two classification problems, a subset of ImageNet and CIFAR-10. For both, we construct large candidate datasets without any explicit human annotations, that only contain 10%-50% correctly labeled examples per class. We show that training on the candidate examples and regularizing through early stopping gives higher test performance for both problems than when training on the original, clean data. This is possible because the candidate datasets contain a huge number of clean examples, and, as we show in this paper, the noise generated through the label collection process is not nearly as adversarial for learning as the noise generated by randomly flipping labels.
Deep learning systems are known to be vulnerable to adversarial examples. In particular, query-based black-box attacks do not require knowledge of the deep learning model, but can compute adversarial examples over the network by submitting queries and inspecting returns. Recent work largely improves the efficiency of those attacks, demonstrating their practicality on today's ML-as-a-service platforms. We propose Blacklight, a new defense against query-based black-box adversarial attacks. The fundamental insight driving our design is that, to compute adversarial examples, these attacks perform iterative optimization over the network, producing image queries highly similar in the input space. Blacklight detects query-based black-box attacks by detecting highly similar queries, using an efficient similarity engine operating on probabilistic content fingerprints. We evaluate Blacklight against eight state-of-the-art attacks, across a variety of models and image classification tasks. Blacklight identifies them all, often after only a handful of queries. By rejecting all detected queries, Blacklight prevents any attack to complete, even when attackers persist to submit queries after account ban or query rejection. Blacklight is also robust against several powerful countermeasures, including an optimal black-box attack that approximates white-box attacks in efficiency. Finally, we illustrate how Blacklight generalizes to other domains like text classification.
There is a warning light for the loss of plant habitats worldwide that entails concerted efforts to conserve plant biodiversity. Thus, plant species classification is of crucial importance to address this environmental challenge. In recent years, there is a considerable increase in the number of studies related to plant taxonomy. While some researchers try to improve their recognition performance using novel approaches, others concentrate on computational optimization for their framework. In addition, a few studies are diving into feature extraction to gain significantly in terms of accuracy. In this paper, we propose an effective method for the leaf recognition problem. In our proposed approach, a leaf goes through some pre-processing to extract its refined color image, vein image, xy-projection histogram, handcrafted shape, texture features, and Fourier descriptors. These attributes are then transformed into a better representation by neural network-based encoders before a support vector machine (SVM) model is utilized to classify different leaves. Overall, our approach performs a state-of-the-art result on the Flavia leaf dataset, achieving the accuracy of 99.58\% on test sets under random 10-fold cross-validation and bypassing the previous methods. We also release our codes (Scripts are available at https://github.com/dinhvietcuong1996/LeafRecognition) for contributing to the research community in the leaf classification problem.
Continual learning (CL) breaks off the one-way training manner and enables a model to adapt to new data, semantics and tasks continuously. However, current CL methods mainly focus on single tasks. Besides, CL models are plagued by catastrophic forgetting and semantic drift since the lack of old data, which often occurs in remote-sensing interpretation due to the intricate fine-grained semantics. In this paper, we propose Continual Panoptic Perception (CPP), a unified continual learning model that leverages multi-task joint learning covering pixel-level classification, instance-level segmentation and image-level perception for universal interpretation in remote sensing images. Concretely, we propose a collaborative cross-modal encoder (CCE) to extract the input image features, which supports pixel classification and caption generation synchronously. To inherit the knowledge from the old model without exemplar memory, we propose a task-interactive knowledge distillation (TKD) method, which leverages cross-modal optimization and task-asymmetric pseudo-labeling (TPL) to alleviate catastrophic forgetting. Furthermore, we also propose a joint optimization mechanism to achieve end-to-end multi-modal panoptic perception. Experimental results on the fine-grained panoptic perception dataset validate the effectiveness of the proposed model, and also prove that joint optimization can boost sub-task CL efficiency with over 13\% relative improvement on panoptic quality.
Nearest neighbor search (NNS) aims to locate the points in high-dimensional space that is closest to the query point. The brute-force approach for finding the nearest neighbor becomes computationally infeasible when the number of points is large. The NNS has multiple applications in medicine, such as searching large medical imaging databases, disease classification, diagnosis, etc. With a focus on medical imaging, this paper proposes DenseLinkSearch an effective and efficient algorithm that searches and retrieves the relevant images from heterogeneous sources of medical images. Towards this, given a medical database, the proposed algorithm builds the index that consists of pre-computed links of each point in the database. The search algorithm utilizes the index to efficiently traverse the database in search of the nearest neighbor. We extensively tested the proposed NNS approach and compared the performance with state-of-the-art NNS approaches on benchmark datasets and our created medical image datasets. The proposed approach outperformed the existing approach in terms of retrieving accurate neighbors and retrieval speed. We also explore the role of medical image feature representation in content-based medical image retrieval tasks. We propose a Transformer-based feature representation technique that outperformed the existing pre-trained Transformer approach on CLEF 2011 medical image retrieval task. The source code of our experiments are available at https://github.com/deepaknlp/DLS.
In this paper, we introduce Random Erasing, a new data augmentation method for training the convolutional neural network (CNN). In training, Random Erasing randomly selects a rectangle region in an image and erases its pixels with random values. In this process, training images with various levels of occlusion are generated, which reduces the risk of over-fitting and makes the model robust to occlusion. Random Erasing is parameter learning free, easy to implement, and can be integrated with most of the CNN-based recognition models. Albeit simple, Random Erasing is complementary to commonly used data augmentation techniques such as random cropping and flipping, and yields consistent improvement over strong baselines in image classification, object detection and person re-identification. Code is available at: https://github.com/zhunzhong07/Random-Erasing.
Modern meta-learning approaches for image classification rely on increasingly deep networks to achieve state-of-the-art performance, making batch normalization an essential component of meta-learning pipelines. However, the hierarchical nature of the meta-learning setting presents several challenges that can render conventional batch normalization ineffective, giving rise to the need to rethink normalization in this setting. We evaluate a range of approaches to batch normalization for meta-learning scenarios, and develop a novel approach that we call TaskNorm. Experiments on fourteen datasets demonstrate that the choice of batch normalization has a dramatic effect on both classification accuracy and training time for both gradient based and gradient-free meta-learning approaches. Importantly, TaskNorm is found to consistently improve performance. Finally, we provide a set of best practices for normalization that will allow fair comparison of meta-learning algorithms.
MetaDL Challenge 2020 focused on image classification tasks in few-shot settings. This paper describes second best submission in the competition. Our meta learning approach modifies the distribution of classes in a latent space produced by a backbone network for each class in order to better follow the Gaussian distribution. After this operation which we call Latent Space Transform algorithm, centers of classes are further aligned in an iterative fashion of the Expectation Maximisation algorithm to utilize information in unlabeled data that are often provided on top of few labelled instances. For this task, we utilize optimal transport mapping using the Sinkhorn algorithm. Our experiments show that this approach outperforms previous works as well as other variants of the algorithm, using K-Nearest Neighbour algorithm, Gaussian Mixture Models, etc.
The limited availability of large image datasets, mainly due to data privacy and differences in acquisition protocols or hardware, is a significant issue in the development of accurate and generalizable machine learning methods in medicine. This is especially the case for Magnetic Resonance (MR) images, where different MR scanners introduce a bias that limits the performance of a machine learning model. We present a novel method that learns to ignore the scanner-related features present in MR images, by introducing specific additional constraints on the latent space. We focus on a real-world classification scenario, where only a small dataset provides images of all classes. Our method \textit{Learn to Ignore (L2I)} outperforms state-of-the-art domain adaptation methods on a multi-site MR dataset for a classification task between multiple sclerosis patients and healthy controls.
Detection of out-of-distribution (OoD) samples in the context of image classification has recently become an area of interest and active study, along with the topic of uncertainty estimation, to which it is closely related. In this paper we explore the task of OoD segmentation, which has been studied less than its classification counterpart and presents additional challenges. Segmentation is a dense prediction task for which the model's outcome for each pixel depends on its surroundings. The receptive field and the reliance on context play a role for distinguishing different classes and, correspondingly, for spotting OoD entities. We introduce MOoSe, an efficient strategy to leverage the various levels of context represented within semantic segmentation models and show that even a simple aggregation of multi-scale representations has consistently positive effects on OoD detection and uncertainty estimation.
In our work, we propose a novel yet simple approach to obtain an adaptive learning rate for gradient-based descent methods on classification tasks. Instead of the traditional approach of selecting adaptive learning rates via the decayed expectation of gradient-based terms, we use the angle between the current gradient and the new gradient: this new gradient is computed from the direction orthogonal to the current gradient, which further helps us in determining a better adaptive learning rate based on angle history, thereby, leading to relatively better accuracy compared to the existing state-of-the-art optimizers. On a wide variety of benchmark datasets with prominent image classification architectures such as ResNet, DenseNet, EfficientNet, and VGG, we find that our method leads to the highest accuracy in most of the datasets. Moreover, we prove that our method is convergent.
Accurate, detailed, and high-frequent bathymetry, coupled with complex semantic content, is crucial for the undermapped shallow seabed areas facing intense climatological and anthropogenic pressures. Current methods exploiting remote sensing images to derive bathymetry or seabed classes mainly exploit non-open data. This lack of openly accessible benchmark archives prevents the wider use of deep learning methods in such applications. To address this issue, in this paper we present the MagicBathyNet, which is a benchmark dataset made up of image patches of Sentinel2, SPOT-6 and aerial imagery, bathymetry in raster format and annotations of seabed classes. MagicBathyNet is then exploited to benchmark state-of-the-art methods in learning-based bathymetry and pixel-based classification. Dataset, pre-trained weights, and code are publicly available at www.magicbathy.eu/magicbathynet.html.