MLP-Mixer has newly appeared as a new challenger against the realm of CNNs and transformer. Despite its simplicity compared to transformer, the concept of channel-mixing MLPs and token-mixing MLPs achieves noticeable performance in visual recognition tasks. Unlike images, point clouds are inherently sparse, unordered and irregular, which limits the direct use of MLP-Mixer for point cloud understanding. In this paper, we propose PointMixer, a universal point set operator that facilitates information sharing among unstructured 3D points. By simply replacing token-mixing MLPs with a softmax function, PointMixer can "mix" features within/between point sets. By doing so, PointMixer can be broadly used in the network as inter-set mixing, intra-set mixing, and pyramid mixing. Extensive experiments show the competitive or superior performance of PointMixer in semantic segmentation, classification, and point reconstruction against transformer-based methods.
Constructing a large-scale labeled dataset in the real world, especially for high-level tasks (eg, Visual Question Answering), can be expensive and time-consuming. In addition, with the ever-growing amounts of data and architecture complexity, Active Learning has become an important aspect of computer vision research. In this work, we address Active Learning in the multi-modal setting of Visual Question Answering (VQA). In light of the multi-modal inputs, image and question, we propose a novel method for effective sample acquisition through the use of ad hoc single-modal branches for each input to leverage its information. Our mutual information based sample acquisition strategy Single-Modal Entropic Measure (SMEM) in addition to our self-distillation technique enables the sample acquisitor to exploit all present modalities and find the most informative samples. Our novel idea is simple to implement, cost-efficient, and readily adaptable to other multi-modal tasks. We confirm our findings on various VQA datasets through state-of-the-art performance by comparing to existing Active Learning baselines.
Videos are a popular media form, where online video streaming has recently gathered much popularity. In this work, we propose a novel method of real-time video stabilization - transforming a shaky video to a stabilized video as if it were stabilized via gimbals in real-time. Our framework is trainable in a self-supervised manner, which does not require data captured with special hardware setups (i.e., two cameras on a stereo rig or additional motion sensors). Our framework consists of a transformation estimator between given frames for global stability adjustments, followed by scene parallax reduction module via spatially smoothed optical flow for further stability. Then, a margin inpainting module fills in the missing margin regions created during stabilization to reduce the amount of post-cropping. These sequential steps reduce distortion and margin cropping to a minimum while enhancing stability. Hence, our approach outperforms state-of-the-art real-time video stabilization methods as well as offline methods that require camera trajectory optimization. Our method procedure takes approximately 24.3 ms yielding 41 fps regardless of resolution (e.g., 480p or 1080p).
Estimating the motion of the camera together with the 3D structure of the scene from a monocular vision system is a complex task that often relies on the so-called scene rigidity assumption. When observing a dynamic environment, this assumption is violated which leads to an ambiguity between the ego-motion of the camera and the motion of the objects. To solve this problem, we present a self-supervised learning framework for 3D object motion field estimation from monocular videos. Our contributions are two-fold. First, we propose a two-stage projection pipeline to explicitly disentangle the camera ego-motion and the object motions with dynamics attention module, called DAM. Specifically, we design an integrated motion model that estimates the motion of the camera and object in the first and second warping stages, respectively, controlled by the attention module through a shared motion encoder. Second, we propose an object motion field estimation through contrastive sample consensus, called CSAC, taking advantage of weak semantic prior (bounding box from an object detector) and geometric constraints (each object respects the rigid body motion model). Experiments on KITTI, Cityscapes, and Waymo Open Dataset demonstrate the relevance of our approach and show that our method outperforms state-of-the-art algorithms for the tasks of self-supervised monocular depth estimation, object motion segmentation, monocular scene flow estimation, and visual odometry.
Convolutional Neural Networks (CNNs) have become the de facto gold standard in computer vision applications in the past years. Recently, however, new model architectures have been proposed challenging the status quo. The Vision Transformer (ViT) relies solely on attention modules, while the MLP-Mixer architecture substitutes the self-attention modules with Multi-Layer Perceptrons (MLPs). Despite their great success, CNNs have been widely known to be vulnerable to adversarial attacks, causing serious concerns for security-sensitive applications. Thus, it is critical for the community to know whether the newly proposed ViT and MLP-Mixer are also vulnerable to adversarial attacks. To this end, we empirically evaluate their adversarial robustness under several adversarial attack setups and benchmark them against the widely used CNNs. Overall, we find that the two architectures, especially ViT, are more robust than their CNN models. Using a toy example, we also provide empirical evidence that the lower adversarial robustness of CNNs can be partially attributed to their shift-invariant property. Our frequency analysis suggests that the most robust ViT architectures tend to rely more on low-frequency features compared with CNNs. Additionally, we have an intriguing finding that MLP-Mixer is extremely vulnerable to universal adversarial perturbations.
Unsupervised domain adaptation (UDA) for semantic segmentation has been attracting attention recently, as it could be beneficial for various label-scarce real-world scenarios (e.g., robot control, autonomous driving, medical imaging, etc.). Despite the significant progress in this field, current works mainly focus on a single-source single-target setting, which cannot handle more practical settings of multiple targets or even unseen targets. In this paper, we investigate open compound domain adaptation (OCDA), which deals with mixed and novel situations at the same time, for semantic segmentation. We present a novel framework based on three main design principles: discover, hallucinate, and adapt. The scheme first clusters compound target data based on style, discovering multiple latent domains (discover). Then, it hallucinates multiple latent target domains in source by using image-translation (hallucinate). This step ensures the latent domains in the source and the target to be paired. Finally, target-to-source alignment is learned separately between domains (adapt). In high-level, our solution replaces a hard OCDA problem with much easier multiple UDA problems. We evaluate our solution on standard benchmark GTA to C-driving, and achieved new state-of-the-art results.
A common problem in the task of human-object interaction (HOI) detection is that numerous HOI classes have only a small number of labeled examples, resulting in training sets with a long-tailed distribution. The lack of positive labels can lead to low classification accuracy for these classes. Towards addressing this issue, we observe that there exist natural correlations and anti-correlations among human-object interactions. In this paper, we model the correlations as action co-occurrence matrices and present techniques to learn these priors and leverage them for more effective training, especially on rare classes. The efficacy of our approach is demonstrated experimentally, where the performance of our approach consistently improves over the state-of-the-art methods on both of the two leading HOI detection benchmark datasets, HICO-Det and V-COCO.
Advances in deep learning recognition have led to accurate object detection with 2D images. However, these 2D perception methods are insufficient for complete 3D world information. Concurrently, advanced 3D shape estimation approaches focus on the shape itself, without considering metric scale. These methods cannot determine the accurate location and orientation of objects. To tackle this problem, we propose a framework that jointly estimates a metric scale shape and pose from a single RGB image. Our framework has two branches: the Metric Scale Object Shape branch (MSOS) and the Normalized Object Coordinate Space branch (NOCS). The MSOS branch estimates the metric scale shape observed in the camera coordinates. The NOCS branch predicts the normalized object coordinate space (NOCS) map and performs similarity transformation with the rendered depth map from a predicted metric scale mesh to obtain 6d pose and size. Additionally, we introduce the Normalized Object Center Estimation (NOCE) to estimate the geometrically aligned distance from the camera to the object center. We validated our method on both synthetic and real-world datasets to evaluate category-level object pose and shape.
To reconstruct a 3D scene from a set of calibrated views, traditional multi-view stereo techniques rely on two distinct stages: local depth maps computation and global depth maps fusion. Recent studies concentrate on deep neural architectures for depth estimation by using conventional depth fusion method or direct 3D reconstruction network by regressing Truncated Signed Distance Function (TSDF). In this paper, we advocate that replicating the traditional two stages framework with deep neural networks improves both the interpretability and the accuracy of the results. As mentioned, our network operates in two steps: 1) the local computation of the local depth maps with a deep MVS technique, and, 2) the depth maps and images' features fusion to build a single TSDF volume. In order to improve the matching performance between images acquired from very different viewpoints (e.g., large-baseline and rotations), we introduce a rotation-invariant 3D convolution kernel called PosedConv. The effectiveness of the proposed architecture is underlined via a large series of experiments conducted on the ScanNet dataset where our approach compares favorably against both traditional and deep learning techniques.