Abstract:This study investigates the application and performance of the Segment Anything Model 2 (SAM2) in the challenging task of video camouflaged object segmentation (VCOS). VCOS involves detecting objects that blend seamlessly in the surroundings for videos, due to similar colors and textures, poor light conditions, etc. Compared to the objects in normal scenes, camouflaged objects are much more difficult to detect. SAM2, a video foundation model, has shown potential in various tasks. But its effectiveness in dynamic camouflaged scenarios remains under-explored. This study presents a comprehensive study on SAM2's ability in VCOS. First, we assess SAM2's performance on camouflaged video datasets using different models and prompts (click, box, and mask). Second, we explore the integration of SAM2 with existing multimodal large language models (MLLMs) and VCOS methods. Third, we specifically adapt SAM2 by fine-tuning it on the video camouflaged dataset. Our comprehensive experiments demonstrate that SAM2 has excellent zero-shot ability of detecting camouflaged objects in videos. We also show that this ability could be further improved by specifically adjusting SAM2's parameters for VCOS. The code will be available at https://github.com/zhoustan/SAM2-VCOS
Abstract:Neural networks achieve state-of-the-art performance in many supervised learning tasks when the training data distribution matches the test data distribution. However, their performance drops significantly under domain (covariate) shift, a prevalent issue in medical image segmentation due to varying acquisition settings across different scanner models and protocols. Recently, foundational models (FMs) trained on large datasets have gained attention for their ability to be adapted for downstream tasks and achieve state-of-the-art performance with excellent generalization capabilities on natural images. However, their effectiveness in medical image segmentation remains underexplored. In this paper, we investigate the domain generalization performance of various FMs, including DinoV2, SAM, MedSAM, and MAE, when fine-tuned using various parameter-efficient fine-tuning (PEFT) techniques such as Ladder and Rein (+LoRA) and decoder heads. We introduce a novel decode head architecture, HQHSAM, which simply integrates elements from two state-of-the-art decoder heads, HSAM and HQSAM, to enhance segmentation performance. Our extensive experiments on multiple datasets, encompassing various anatomies and modalities, reveal that FMs, particularly with the HQHSAM decode head, improve domain generalization for medical image segmentation. Moreover, we found that the effectiveness of PEFT techniques varies across different FMs. These findings underscore the potential of FMs to enhance the domain generalization performance of neural networks in medical image segmentation across diverse clinical settings, providing a solid foundation for future research. Code and models are available for research purposes at \url{https://github.com/kerem-cekmeceli/Foundation-Models-for-Medical-Imagery}.
Abstract:Image segmentation is a long-standing challenge in computer vision, studied continuously over several decades, as evidenced by seminal algorithms such as N-Cut, FCN, and MaskFormer. With the advent of foundation models (FMs), contemporary segmentation methodologies have embarked on a new epoch by either adapting FMs (e.g., CLIP, Stable Diffusion, DINO) for image segmentation or developing dedicated segmentation foundation models (e.g., SAM). These approaches not only deliver superior segmentation performance, but also herald newfound segmentation capabilities previously unseen in deep learning context. However, current research in image segmentation lacks a detailed analysis of distinct characteristics, challenges, and solutions associated with these advancements. This survey seeks to fill this gap by providing a thorough review of cutting-edge research centered around FM-driven image segmentation. We investigate two basic lines of research -- generic image segmentation (i.e., semantic segmentation, instance segmentation, panoptic segmentation), and promptable image segmentation (i.e., interactive segmentation, referring segmentation, few-shot segmentation) -- by delineating their respective task settings, background concepts, and key challenges. Furthermore, we provide insights into the emergence of segmentation knowledge from FMs like CLIP, Stable Diffusion, and DINO. An exhaustive overview of over 300 segmentation approaches is provided to encapsulate the breadth of current research efforts. Subsequently, we engage in a discussion of open issues and potential avenues for future research. We envisage that this fresh, comprehensive, and systematic survey catalyzes the evolution of advanced image segmentation systems.
Abstract:3D Gaussian Splatting (3DGS) has become the de facto method of 3D representation in many vision tasks. This calls for the 3D understanding directly in this representation space. To facilitate the research in this direction, we first build a large-scale dataset of 3DGS using the commonly used ShapeNet and ModelNet datasets. Our dataset ShapeSplat consists of 65K objects from 87 unique categories, whose labels are in accordance with the respective datasets. The creation of this dataset utilized the compute equivalent of 2 GPU years on a TITAN XP GPU. We utilize our dataset for unsupervised pretraining and supervised finetuning for classification and segmentation tasks. To this end, we introduce \textbf{\textit{Gaussian-MAE}}, which highlights the unique benefits of representation learning from Gaussian parameters. Through exhaustive experiments, we provide several valuable insights. In particular, we show that (1) the distribution of the optimized GS centroids significantly differs from the uniformly sampled point cloud (used for initialization) counterpart; (2) this change in distribution results in degradation in classification but improvement in segmentation tasks when using only the centroids; (3) to leverage additional Gaussian parameters, we propose Gaussian feature grouping in a normalized feature space, along with splats pooling layer, offering a tailored solution to effectively group and embed similar Gaussians, which leads to notable improvement in finetuning tasks.
Abstract:Neural implicit functions have demonstrated significant importance in various areas such as computer vision, graphics. Their advantages include the ability to represent complex shapes and scenes with high fidelity, smooth interpolation capabilities, and continuous representations. Despite these benefits, the development and analysis of implicit functions have been limited by the lack of comprehensive datasets and the substantial computational resources required for their implementation and evaluation. To address these challenges, we introduce "Implicit-Zoo": a large-scale dataset requiring thousands of GPU training days designed to facilitate research and development in this field. Our dataset includes diverse 2D and 3D scenes, such as CIFAR-10, ImageNet-1K, and Cityscapes for 2D image tasks, and the OmniObject3D dataset for 3D vision tasks. We ensure high quality through strict checks, refining or filtering out low-quality data. Using Implicit-Zoo, we showcase two immediate benefits as it enables to: (1) learn token locations for transformer models; (2) directly regress 3D cameras poses of 2D images with respect to NeRF models. This in turn leads to an improved performance in all three task of image classification, semantic segmentation, and 3D pose regression, thereby unlocking new avenues for research.
Abstract:Implicit functions such as Neural Radiance Fields (NeRFs), occupancy networks, and signed distance functions (SDFs) have become pivotal in computer vision for reconstructing detailed object shapes from sparse views. Achieving optimal performance with these models can be challenging due to the extreme sparsity of inputs and distribution shifts induced by data corruptions. To this end, large, noise-free synthetic datasets can serve as shape priors to help models fill in gaps, but the resulting reconstructions must be approached with caution. Uncertainty estimation is crucial for assessing the quality of these reconstructions, particularly in identifying areas where the model is uncertain about the parts it has inferred from the prior. In this paper, we introduce Dropsembles, a novel method for uncertainty estimation in tuned implicit functions. We demonstrate the efficacy of our approach through a series of experiments, starting with toy examples and progressing to a real-world scenario. Specifically, we train a Convolutional Occupancy Network on synthetic anatomical data and test it on low-resolution MRI segmentations of the lumbar spine. Our results show that Dropsembles achieve the accuracy and calibration levels of deep ensembles but with significantly less computational cost.
Abstract:Localizing oneself during endoscopic procedures can be problematic due to the lack of distinguishable textures and landmarks, as well as difficulties due to the endoscopic device such as a limited field of view and challenging lighting conditions. Expert knowledge shaped by years of experience is required for localization within the human body during endoscopic procedures. In this work, we present a deep learning method based on anatomy recognition, that constructs a surgical path in an unsupervised manner from surgical videos, modelling relative location and variations due to different viewing angles. At inference time, the model can map an unseen video's frames on the path and estimate the viewing angle, aiming to provide guidance, for instance, to reach a particular destination. We test the method on a dataset consisting of surgical videos of transsphenoidal adenomectomies, as well as on a synthetic dataset. An online tool that lets researchers upload their surgical videos to obtain anatomy detections and the weights of the trained YOLOv7 model are available at: https://surgicalvision.bmic.ethz.ch.
Abstract:Manifold learning flows are a class of generative modelling techniques that assume a low-dimensional manifold description of the data. The embedding of such a manifold into the high-dimensional space of the data is achieved via learnable invertible transformations. Therefore, once the manifold is properly aligned via a reconstruction loss, the probability density is tractable on the manifold and maximum likelihood can be used to optimize the network parameters. Naturally, the lower-dimensional representation of the data requires an injective-mapping. Recent approaches were able to enforce that the density aligns with the modelled manifold, while efficiently calculating the density volume-change term when embedding to the higher-dimensional space. However, unless the injective-mapping is analytically predefined, the learned manifold is not necessarily an efficient representation of the data. Namely, the latent dimensions of such models frequently learn an entangled intrinsic basis, with degenerate information being stored in each dimension. Alternatively, if a locally orthogonal and/or sparse basis is to be learned, here coined canonical intrinsic basis, it can serve in learning a more compact latent space representation. Toward this end, we propose a canonical manifold learning flow method, where a novel optimization objective enforces the transformation matrix to have few prominent and non-degenerate basis functions. We demonstrate that by minimizing the off-diagonal manifold metric elements $\ell_1$-norm, we can achieve such a basis, which is simultaneously sparse and/or orthogonal. Canonical manifold flow yields a more efficient use of the latent space, automatically generating fewer prominent and distinct dimensions to represent data, and a better approximation of target distributions than other manifold flow methods in most experiments we conducted, resulting in lower FID scores.
Abstract:In human-AI collaboration systems for critical applications, in order to ensure minimal error, users should set an operating point based on model confidence to determine when the decision should be delegated to human experts. Samples for which model confidence is lower than the operating point would be manually analysed by experts to avoid mistakes. Such systems can become truly useful only if they consider two aspects: models should be confident only for samples for which they are accurate, and the number of samples delegated to experts should be minimized. The latter aspect is especially crucial for applications where available expert time is limited and expensive, such as healthcare. The trade-off between the model accuracy and the number of samples delegated to experts can be represented by a curve that is similar to an ROC curve, which we refer to as confidence operating characteristic (COC) curve. In this paper, we argue that deep neural networks should be trained by taking into account both accuracy and expert load and, to that end, propose a new complementary loss function for classification that maximizes the area under this COC curve. This promotes simultaneously the increase in network accuracy and the reduction in number of samples delegated to humans. We perform experiments on multiple computer vision and medical image datasets for classification. Our results demonstrate that the proposed loss improves classification accuracy and delegates less number of decisions to experts, achieves better out-of-distribution samples detection and on par calibration performance compared to existing loss functions.
Abstract:Predictive variability due to data ambiguities has typically been addressed via construction of dedicated models with built-in probabilistic capabilities that are trained to predict uncertainty estimates as variables of interest. These approaches require distinct architectural components and training mechanisms, may include restrictive assumptions and exhibit overconfidence, i.e., high confidence in imprecise predictions. In this work, we propose a post-hoc sampling strategy for estimating predictive uncertainty accounting for data ambiguity. The method can generate different plausible outputs for a given input and does not assume parametric forms of predictive distributions. It is architecture agnostic and can be applied to any feed-forward deterministic network without changes to the architecture or training procedure. Experiments on regression tasks on imaging and non-imaging input data show the method's ability to generate diverse and multi-modal predictive distributions, and a desirable correlation of the estimated uncertainty with the prediction error.