A household robot should be able to navigate to target locations without requiring users to first annotate everything in their home. Current approaches to this object navigation challenge do not test on real robots and rely on expensive semantically labeled 3D meshes. In this work, our aim is an agent that builds self-supervised models of the world via exploration, the same as a child might. We propose an end-to-end self-supervised embodied agent that leverages exploration to train a semantic segmentation model of 3D objects, and uses those representations to learn an object navigation policy purely from self-labeled 3D meshes. The key insight is that embodied agents can leverage location consistency as a supervision signal - collecting images from different views/angles and applying contrastive learning to fine-tune a semantic segmentation model. In our experiments, we observe that our framework performs better than other self-supervised baselines and competitively with supervised baselines, in both simulation and when deployed in real houses.
Most existing Image Restoration (IR) models are task-specific, which can not be generalized to different degradation operators. In this work, we propose the Denoising Diffusion Null-Space Model (DDNM), a novel zero-shot framework for arbitrary linear IR problems, including but not limited to image super-resolution, colorization, inpainting, compressed sensing, and deblurring. DDNM only needs a pre-trained off-the-shelf diffusion model as the generative prior, without any extra training or network modifications. By refining only the null-space contents during the reverse diffusion process, we can yield diverse results satisfying both data consistency and realness. We further propose an enhanced and robust version, dubbed DDNM+, to support noisy restoration and improve restoration quality for hard tasks. Our experiments on several IR tasks reveal that DDNM outperforms other state-of-the-art zero-shot IR methods. We also demonstrate that DDNM+ can solve complex real-world applications, e.g., old photo restoration.
Consistency and realness have always been the two critical issues of image super-resolution. While the realness has been dramatically improved with the use of GAN prior, the state-of-the-art methods still suffer inconsistencies in local structures and colors (e.g., tooth and eyes). In this paper, we show that these inconsistencies can be analytically eliminated by learning only the null-space component while fixing the range-space part. Further, we design a pooling-based decomposition (PD), a universal range-null space decomposition for super-resolution tasks, which is concise, fast, and parameter-free. PD can be easily applied to state-of-the-art GAN Prior based SR methods to eliminate their inconsistencies, neither compromising the realness nor bringing extra parameters or computational costs. Besides, our ablation studies reveal that PD can replace pixel-wise losses for training and achieve better generalization performance when facing unseen downsamplings or even real-world degradation. Experiments show that the use of PD refreshes state-of-the-art SR performance and speeds up the convergence of training up to 2~10 times.
Complementary Labels Learning (CLL) arises in many real-world tasks such as private questions classification and online learning, which aims to alleviate the annotation cost compared with standard supervised learning. Unfortunately, most previous CLL algorithms were in a stable environment rather than an open and dynamic scenarios, where data collected from unseen augmented classes in the training process might emerge in the testing phase. In this paper, we propose a novel problem setting called Complementary Labels Learning with Augmented Classes (CLLAC), which brings the challenge that classifiers trained by complementary labels should not only be able to classify the instances from observed classes accurately, but also recognize the instance from the Augmented Classes in the testing phase. Specifically, by using unlabeled data, we propose an unbiased estimator of classification risk for CLLAC, which is guaranteed to be provably consistent. Moreover, we provide generalization error bound for proposed method which shows that the optimal parametric convergence rate is achieved for estimation error. Finally, the experimental results on several benchmark datasets verify the effectiveness of the proposed method.
Scene text recognition (STR) enables computers to recognize and read the text in various real-world scenes. Recent STR models benefit from taking linguistic information in addition to visual cues into consideration. We propose a novel Masked Vision-Language Transformers (MVLT) to capture both the explicit and the implicit linguistic information. Our encoder is a Vision Transformer, and our decoder is a multi-modal Transformer. MVLT is trained in two stages: in the first stage, we design a STR-tailored pretraining method based on a masking strategy; in the second stage, we fine-tune our model and adopt an iterative correction method to improve the performance. MVLT attains superior results compared to state-of-the-art STR models on several benchmarks. Our code and model are available at https://github.com/onealwj/MVLT.
In this paper, we propose multi-agent automated machine learning (MA2ML) with the aim to effectively handle joint optimization of modules in automated machine learning (AutoML). MA2ML takes each machine learning module, such as data augmentation (AUG), neural architecture search (NAS), or hyper-parameters (HPO), as an agent and the final performance as the reward, to formulate a multi-agent reinforcement learning problem. MA2ML explicitly assigns credit to each agent according to its marginal contribution to enhance cooperation among modules, and incorporates off-policy learning to improve search efficiency. Theoretically, MA2ML guarantees monotonic improvement of joint optimization. Extensive experiments show that MA2ML yields the state-of-the-art top-1 accuracy on ImageNet under constraints of computational cost, e.g., $79.7\%/80.5\%$ with FLOPs fewer than 600M/800M. Extensive ablation studies verify the benefits of credit assignment and off-policy learning of MA2ML.
Probabilistic 3D point cloud registration methods have shown competitive performance in overcoming noise, outliers, and density variations. However, registering point cloud pairs in the case of partial overlap is still a challenge. This paper proposes a novel overlap-guided probabilistic registration approach that computes the optimal transformation from matched Gaussian Mixture Model (GMM) parameters. We reformulate the registration problem as the problem of aligning two Gaussian mixtures such that a statistical discrepancy measure between the two corresponding mixtures is minimized. We introduce a Transformer-based detection module to detect overlapping regions, and represent the input point clouds using GMMs by guiding their alignment through overlap scores computed by this detection module. Experiments show that our method achieves superior registration accuracy and efficiency than state-of-the-art methods when handling point clouds with partial overlap and different densities on synthetic and real-world datasets. https://github.com/gfmei/ogmm
Unsupervised learning on 3D point clouds has undergone a rapid evolution, especially thanks to data augmentation-based contrastive methods. However, data augmentation is not ideal as it requires a careful selection of the type of augmentations to perform, which in turn can affect the geometric and semantic information learned by the network during self-training. To overcome this issue, we propose an augmentation-free unsupervised approach for point clouds to learn transferable point-level features via soft clustering, named SoftClu. SoftClu assumes that the points belonging to a cluster should be close to each other in both geometric and feature spaces. This differs from typical contrastive learning, which builds similar representations for a whole point cloud and its augmented versions. We exploit the affiliation of points to their clusters as a proxy to enable self-training through a pseudo-label prediction task. Under the constraint that these pseudo-labels induce the equipartition of the point cloud, we cast SoftClu as an optimal transport problem. We formulate an unsupervised loss to minimize the standard cross-entropy between pseudo-labels and predicted labels. Experiments on downstream applications, such as 3D object classification, part segmentation, and semantic segmentation, show the effectiveness of our framework in outperforming state-of-the-art techniques.
3D visual grounding aims to find the objects within point clouds mentioned by free-form natural language descriptions with rich semantic components. However, existing methods either extract the sentence-level features coupling all words, or focus more on object names, which would lose the word-level information or neglect other attributes. To alleviate this issue, we present EDA that Explicitly Decouples the textual attributes in a sentence and conducts Dense Alignment between such fine-grained language and point cloud objects. Specifically, we first propose a text decoupling module to produce textual features for every semantic component. Then, we design two losses to supervise the dense matching between two modalities: the textual position alignment and object semantic alignment. On top of that, we further introduce two new visual grounding tasks, locating objects without object names and locating auxiliary objects referenced in the descriptions, both of which can thoroughly evaluate the model's dense alignment capacity. Through experiments, we achieve state-of-the-art performance on two widely-adopted visual grounding datasets , ScanRefer and SR3D/NR3D, and obtain absolute leadership on our two newly-proposed tasks. The code will be available at https://github.com/yanmin-wu/EDA.