The recent Segment Anything Model (SAM) has demonstrated remarkable zero-shot capability and flexible geometric prompting in general image segmentation. However, SAM often struggles when handling various unconventional images, such as aerial, medical, and non-RGB images. This paper presents CAT-SAM, a ConditionAl Tuning network that adapts SAM toward various unconventional target tasks with just few-shot target samples. CAT-SAM freezes the entire SAM and adapts its mask decoder and image encoder simultaneously with a small number of learnable parameters. The core design is a prompt bridge structure that enables decoder-conditioned joint tuning of the heavyweight image encoder and the lightweight mask decoder. The bridging maps the prompt token of the mask decoder to the image encoder, fostering synergic adaptation of the encoder and the decoder with mutual benefits. We develop two representative tuning strategies for the image encoder which leads to two CAT-SAM variants: one injecting learnable prompt tokens in the input space and the other inserting lightweight adapter networks. Extensive experiments over 11 unconventional tasks show that both CAT-SAM variants achieve superior target segmentation performance consistently even under the very challenging one-shot adaptation setup. Project page: \url{https://xiaoaoran.github.io/projects/CAT-SAM}
The transformation model is an essential component of any deformable image registration approach. It provides a representation of physical deformations between images, thereby defining the range and realism of registrations that can be found. Two types of transformation models have emerged as popular choices: B-spline models and mesh models. Although both models have been investigated in detail, a direct comparison has not yet been made, since the models are optimized using very different optimization methods in practice. B-spline models are predominantly optimized using gradient-descent methods, while mesh models are typically optimized using finite-element method solvers or evolutionary algorithms. Multi-objective optimization methods, which aim to find a diverse set of high-quality trade-off registrations, are increasingly acknowledged to be important in deformable image registration. Since these methods search for a diverse set of registrations, they can provide a more complete picture of the capabilities of different transformation models, making them suitable for a comparison of models. In this work, we conduct the first direct comparison between B-spline and mesh transformation models, by optimizing both models with the same state-of-the-art multi-objective optimization method, the Multi-Objective Real-Valued Gene-pool Optimal Mixing Evolutionary Algorithm (MO-RV-GOMEA). The combination with B-spline transformation models, moreover, is novel. We experimentally compare both models on two different registration problems that are both based on pelvic CT scans of cervical cancer patients, featuring large deformations. Our results, on three cervical cancer patients, indicate that the choice of transformation model can have a profound impact on the diversity and quality of achieved registration outcomes.
While branching network structures abound in nature, their objective analysis is more difficult than expected because existing quantitative methods often rely on the subjective judgment of branch structures. This problem is particularly pronounced when dealing with images comprising discrete particles. Here we propose an objective framework for quantitative analysis of branching networks by introducing the mathematical definitions for internal and external structures based on topological data analysis, specifically, persistent homology. We compare persistence diagrams constructed from images with and without plots on the convex hull. The unchanged points in the two diagrams are the internal structures and the difference between the two diagrams is the external structures. We construct a mathematical theory for our method and show that the internal structures have a monotonicity relationship with respect to the plots on the convex hull, while the external structures do not. This is the phenomenon related to the resolution of the image. Our method can be applied to a wide range of branch structures in biology, enabling objective analysis of numbers, spatial distributions, sizes, and more. Additionally, our method has the potential to be combined with other tools in topological data analysis, such as the generalized persistence landscape.
Multimodal Large Language Models (MLLMs) have recently shown remarkable perceptual capability in answering visual questions, however, little is known about the limits of their perception. In particular, while prior works have provided anecdotal evidence of MLLMs' sensitivity to object size, this phenomenon and its underlying causes have not been explored comprehensively. In this work, we quantitatively study the perception of small visual objects in several state-of-the-art MLLMs and reveal a pervasive limitation in answering questions about small objects in images. Next, we identify four independent factors that can contribute to this limitation -- object quality, size, distractors, and location -- and conduct controlled intervention studies to measure the effect of each factor on MLLMs' perception. In particular, we find that lower object quality and smaller object size can both independently reduce MLLMs' ability to answer visual questions. More surprisingly, we find that the location of the object in the image and the presence of visual distractors can also significantly reduce MLLMs' question answering accuracy. Our study provides a better understanding of the perceptual limitation of MLLMs and contributes new evaluation protocols for analyzing the perception of future MLLMs. To facilitate further investigations, we release our code and data.
Atmospheric turbulence, a common phenomenon in daily life, is primarily caused by the uneven heating of the Earth's surface. This phenomenon results in distorted and blurred acquired images or videos and can significantly impact downstream vision tasks, particularly those that rely on capturing clear, stable images or videos from outdoor environments, such as accurately detecting or recognizing objects. Therefore, people have proposed ways to simulate atmospheric turbulence and designed effective deep learning-based methods to remove the atmospheric turbulence effect. However, these synthesized turbulent images can not cover all the range of real-world turbulence effects. Though the models have achieved great performance for synthetic scenarios, there always exists a performance drop when applied to real-world cases. Moreover, reducing real-world turbulence is a more challenging task as there are no clean ground truth counterparts provided to the models during training. In this paper, we propose a real-world atmospheric turbulence mitigation model under a domain adaptation framework, which links the supervised simulated atmospheric turbulence correction with the unsupervised real-world atmospheric turbulence correction. We will show our proposed method enhances performance in real-world atmospheric turbulence scenarios, improving both image quality and downstream vision tasks.
Visually-conditioned language models (VLMs) have seen growing adoption in applications such as visual dialogue, scene understanding, and robotic task planning; adoption that has fueled a wealth of new models such as LLaVa, InstructBLIP, and PaLI-3. Despite the volume of new releases, key design decisions around image preprocessing, architecture, and optimization are under-explored, making it challenging to understand what factors account for model performance $-$ a challenge further complicated by the lack of objective, consistent evaluations. To address these gaps, we first compile a suite of standardized evaluations spanning visual question answering, object localization from language, and targeted challenge sets that probe properties such as hallucination; evaluations that provide calibrated, fine-grained insight into a VLM's capabilities. Second, we rigorously investigate VLMs along key design axes, including pretrained visual representations and quantifying the tradeoffs of using base vs. instruct-tuned language models, amongst others. We couple our analysis with three resource contributions: (1) a unified framework for evaluating VLMs, (2) optimized, flexible code for VLM training, and (3) checkpoints for all models, including a family of VLMs at the 7-13B scale that strictly outperform InstructBLIP and LLaVa v1.5, the state-of-the-art in open-source VLMs.
Recent work has shown the defense of 01 loss sign activation neural networks against image classification adversarial attacks. A public challenge to attack the models on CIFAR10 dataset remains undefeated. We ask the following question in this study: are 01 loss sign activation neural networks hard to deceive with a popular black box text adversarial attack program called TextFooler? We study this question on four popular text classification datasets: IMDB reviews, Yelp reviews, MR sentiment classification, and AG news classification. We find that our 01 loss sign activation network is much harder to attack with TextFooler compared to sigmoid activation cross entropy and binary neural networks. We also study a 01 loss sign activation convolutional neural network with a novel global pooling step specific to sign activation networks. With this new variation we see a significant gain in adversarial accuracy rendering TextFooler practically useless against it. We make our code freely available at \url{https://github.com/zero-one-loss/wordcnn01} and \url{https://github.com/xyzacademic/mlp01example}. Our work here suggests that 01 loss sign activation networks could be further developed to create fool proof models against text adversarial attacks.
There is a significant interest in exploring non-linear associations among multiple images derived from diverse imaging modalities. While there is a growing literature on image-on-image regression to delineate predictive inference of an image based on multiple images, existing approaches have limitations in efficiently borrowing information between multiple imaging modalities in the prediction of an image. Building on the literature of Variational Auto Encoders (VAEs), this article proposes a novel approach, referred to as Integrative Variational Autoencoder (\texttt{InVA}) method, which borrows information from multiple images obtained from different sources to draw predictive inference of an image. The proposed approach captures complex non-linear association between the outcome image and input images, while allowing rapid computation. Numerical results demonstrate substantial advantages of \texttt{InVA} over VAEs, which typically do not allow borrowing information between input images. The proposed framework offers highly accurate predictive inferences for costly positron emission topography (PET) from multiple measures of cortical structure in human brain scans readily available from magnetic resonance imaging (MRI).
Semantic communication is focused on optimizing the exchange of information by transmitting only the most relevant data required to convey the intended message to the receiver and achieve the desired communication goal. For example, if we consider images as the information and the goal of the communication is object detection at the receiver side, the semantic of information would be the objects in each image. Therefore, by only transferring the semantics of images we can achieve the communication goal. In this paper, we propose a design framework for implementing semantic-aware and goal-oriented communication of images. To achieve this, we first define the baseline problem as a set of mathematical problems that can be optimized to improve the efficiency and effectiveness of the communication system. We consider two scenarios in which either the data rate or the error at the receiver is the limiting constraint. Our proposed system model and solution is inspired by the concept of auto-encoders, where the encoder and the decoder are respectively implemented at the transmitter and receiver to extract semantic information for specific object detection goals. Our numerical results validate the proposed design framework to achieve low error or near-optimal in a goal-oriented communication system while reducing the amount of data transfers.
In this paper, we propose a new visual navigation method based on a single RGB perspective camera. Using the Visual Teach & Repeat (VT&R) methodology, the robot acquires a visual trajectory consisting of multiple subgoal images in the teaching step. In the repeat step, we propose two network architectures, namely ViewNet and VelocityNet. The combination of the two networks allows the robot to follow the visual trajectory. ViewNet is trained to generate a future image based on the current view and the velocity command. The generated future image is combined with the subgoal image for training VelocityNet. We develop an offline Model Predictive Control (MPC) policy within VelocityNet with the dual goals of (1) reducing the difference between current and subgoal images and (2) ensuring smooth trajectories by mitigating velocity discontinuities. Offline training conserves computational resources, making it a more suitable option for scenarios with limited computational capabilities, such as embedded systems. We validate our experiments in a simulation environment, demonstrating that our model can effectively minimize the metric error between real and played trajectories.