Image resizing operation is a fundamental preprocessing module in modern computer vision. Throughout the deep learning revolution, researchers have overlooked the potential of alternative resizing methods beyond the commonly used resizers that are readily available, such as nearest-neighbors, bilinear, and bicubic. The key question of our interest is whether the front-end resizer affects the performance of deep vision models? In this paper, we present an extremely lightweight multilayer Laplacian resizer with only a handful of trainable parameters, dubbed MULLER resizer. MULLER has a bandpass nature in that it learns to boost details in certain frequency subbands that benefit the downstream recognition models. We show that MULLER can be easily plugged into various training pipelines, and it effectively boosts the performance of the underlying vision task with little to no extra cost. Specifically, we select a state-of-the-art vision Transformer, MaxViT, as the baseline, and show that, if trained with MULLER, MaxViT gains up to 0.6% top-1 accuracy, and meanwhile enjoys 36% inference cost saving to achieve similar top-1 accuracy on ImageNet-1k, as compared to the standard training scheme. Notably, MULLER's performance also scales with model size and training data size such as ImageNet-21k and JFT, and it is widely applicable to multiple vision tasks, including image classification, object detection and segmentation, as well as image quality assessment.
Image-text retrieval (ITR) is a challenging task in the field of multimodal information processing due to the semantic gap between different modalities. In recent years, researchers have made great progress in exploring the accurate alignment between image and text. However, existing works mainly focus on the fine-grained alignment between image regions and sentence fragments, which ignores the guiding significance of context background information. Actually, integrating the local fine-grained information and global context background information can provide more semantic clues for retrieval. In this paper, we propose a novel Hierarchical Graph Alignment Network (HGAN) for image-text retrieval. First, to capture the comprehensive multimodal features, we construct the feature graphs for the image and text modality respectively. Then, a multi-granularity shared space is established with a designed Multi-granularity Feature Aggregation and Rearrangement (MFAR) module, which enhances the semantic corresponding relations between the local and global information, and obtains more accurate feature representations for the image and text modalities. Finally, the ultimate image and text features are further refined through three-level similarity functions to achieve the hierarchical alignment. To justify the proposed model, we perform extensive experiments on MS-COCO and Flickr30K datasets. Experimental results show that the proposed HGAN outperforms the state-of-the-art methods on both datasets, which demonstrates the effectiveness and superiority of our model.
While recent research has made significant progress in speech-driven talking face generation, the quality of the generated video still lags behind that of real recordings. One reason for this is the use of handcrafted intermediate representations like facial landmarks and 3DMM coefficients, which are designed based on human knowledge and are insufficient to precisely describe facial movements. Additionally, these methods require an external pretrained model for extracting these representations, whose performance sets an upper bound on talking face generation. To address these limitations, we propose a novel method called DAE-Talker that leverages data-driven latent representations obtained from a diffusion autoencoder (DAE). DAE contains an image encoder that encodes an image into a latent vector and a DDIM image decoder that reconstructs the image from it. We train our DAE on talking face video frames and then extract their latent representations as the training target for a Conformer-based speech2latent model. This allows DAE-Talker to synthesize full video frames and produce natural head movements that align with the content of speech, rather than relying on a predetermined head pose from a template video. We also introduce pose modelling in speech2latent for pose controllability. Additionally, we propose a novel method for generating continuous video frames with the DDIM image decoder trained on individual frames, eliminating the need for modelling the joint distribution of consecutive frames directly. Our experiments show that DAE-Talker outperforms existing popular methods in lip-sync, video fidelity, and pose naturalness. We also conduct ablation studies to analyze the effectiveness of the proposed techniques and demonstrate the pose controllability of DAE-Talker.
Learning rich data representations from unlabeled data is a key challenge towards applying deep learning algorithms in downstream supervised tasks. Several variants of variational autoencoders have been proposed to learn compact data representaitons by encoding high-dimensional data in a lower dimensional space. Two main classes of VAEs methods may be distinguished depending on the characteristics of the meta-priors that are enforced in the representation learning step. The first class of methods derives a continuous encoding by assuming a static prior distribution in the latent space. The second class of methods learns instead a discrete latent representation using vector quantization (VQ) along with a codebook. However, both classes of methods suffer from certain challenges, which may lead to suboptimal image reconstruction results. The first class of methods suffers from posterior collapse, whereas the second class of methods suffers from codebook collapse. To address these challenges, we introduce a new VAE variant, termed SC-VAE (sparse coding-based VAE), which integrates sparse coding within variational autoencoder framework. Instead of learning a continuous or discrete latent representation, the proposed method learns a sparse data representation that consists of a linear combination of a small number of learned atoms. The sparse coding problem is solved using a learnable version of the iterative shrinkage thresholding algorithm (ISTA). Experiments on two image datasets demonstrate that our model can achieve improved image reconstruction results compared to state-of-the-art methods. Moreover, the use of learned sparse code vectors allows us to perform downstream task like coarse image segmentation through clustering image patches.
Unsupervised domain adaptation (UDA) for semantic segmentation is a promising task freeing people from heavy annotation work. However, domain discrepancies in low-level image statistics and high-level contexts compromise the segmentation performance over the target domain. A key idea to tackle this problem is to perform both image-level and feature-level adaptation jointly. Unfortunately, there is a lack of such unified approaches for UDA tasks in the existing literature. This paper proposes a novel UDA pipeline for semantic segmentation that unifies image-level and feature-level adaptation. Concretely, for image-level domain shifts, we propose a global photometric alignment module and a global texture alignment module that align images in the source and target domains in terms of image-level properties. For feature-level domain shifts, we perform global manifold alignment by projecting pixel features from both domains onto the feature manifold of the source domain; and we further regularize category centers in the source domain through a category-oriented triplet loss and perform target domain consistency regularization over augmented target domain images. Experimental results demonstrate that our pipeline significantly outperforms previous methods. In the commonly tested GTA5$\rightarrow$Cityscapes task, our proposed method using Deeplab V3+ as the backbone surpasses previous SOTA by 8%, achieving 58.2% in mIoU.
Keypoint detection & descriptors are foundational tech-nologies for computer vision tasks like image matching, 3D reconstruction and visual odometry. Hand-engineered methods like Harris corners, SIFT, and HOG descriptors have been used for decades; more recently, there has been a trend to introduce learning in an attempt to improve keypoint detectors. On inspection however, the results are difficult to interpret; recent learning-based methods employ a vast diversity of experimental setups and design choices: empirical results are often reported using different backbones, protocols, datasets, types of supervisions or tasks. Since these differences are often coupled together, it raises a natural question on what makes a good learned keypoint detector. In this work, we revisit the design of existing keypoint detectors by deconstructing their methodologies and identifying the key components. We re-design each component from first-principle and propose Simple Learned Keypoints (SiLK) that is fully-differentiable, lightweight, and flexible. Despite its simplicity, SiLK advances new state-of-the-art on Detection Repeatability and Homography Estimation tasks on HPatches and 3D Point-Cloud Registration task on ScanNet, and achieves competitive performance to state-of-the-art on camera pose estimation in 2022 Image Matching Challenge and ScanNet.
Quality control (QC) has long been considered essential to guarantee the reliability of neuroimaging studies. It is particularly important for fetal brain MRI, where large and unpredictable fetal motion can lead to substantial artifacts in the acquired images. Existing methods for fetal brain quality assessment operate at the \textit{slice} level, and fail to get a comprehensive picture of the quality of an image, that can only be achieved by looking at the \textit{entire} brain volume. In this work, we propose FetMRQC, a machine learning framework for automated image quality assessment tailored to fetal brain MRI, which extracts an ensemble of quality metrics that are then used to predict experts' ratings. Based on the manual ratings of more than 1000 low-resolution stacks acquired across two different institutions, we show that, compared with existing quality metrics, FetMRQC is able to generalize out-of-domain, while being interpretable and data efficient. We also release a novel manual quality rating tool designed to facilitate and optimize quality rating of fetal brain images. Our tool, along with all the code to generate, train and evaluate the model will be released upon acceptance of the paper.
Recent works on personalized text-to-image generation usually learn to bind a special token with specific subjects or styles of a few given images by tuning its embedding through gradient descent. It is natural to question whether we can optimize the textual inversions by only accessing the process of model inference. As only requiring the forward computation to determine the textual inversion retains the benefits of less GPU memory, simple deployment, and secure access for scalable models. In this paper, we introduce a \emph{gradient-free} framework to optimize the continuous textual inversion in an iterative evolutionary strategy. Specifically, we first initialize an appropriate token embedding for textual inversion with the consideration of visual and text vocabulary information. Then, we decompose the optimization of evolutionary strategy into dimension reduction of searching space and non-convex gradient-free optimization in subspace, which significantly accelerates the optimization process with negligible performance loss. Experiments in several applications demonstrate that the performance of text-to-image model equipped with our proposed gradient-free method is comparable to that of gradient-based counterparts with variant GPU/CPU platforms, flexible employment, as well as computational efficiency.
Neuromorphic (event-based) image sensors draw inspiration from the human-retina to create an electronic device that can process visual stimuli in a way that closely resembles its biological counterpart. These sensors process information significantly different than the traditional RGB sensors. Specifically, the sensory information generated by event-based image sensors are orders of magnitude sparser compared to that of RGB sensors. The first generation of neuromorphic image sensors, Dynamic Vision Sensor (DVS), are inspired by the computations confined to the photoreceptors and the first retinal synapse. In this work, we highlight the capability of the second generation of neuromorphic image sensors, Integrated Retinal Functionality in CMOS Image Sensors (IRIS), which aims to mimic full retinal computations from photoreceptors to output of the retina (retinal ganglion cells) for targeted feature-extraction. The feature of choice in this work is Object Motion Sensitivity (OMS) that is processed locally in the IRIS sensor. We study the capability of OMS in solving the ego-motion problem of the event-based cameras. Our results show that OMS can accomplish standard computer vision tasks with similar efficiency to conventional RGB and DVS solutions but offers drastic bandwidth reduction. This cuts the wireless and computing power budgets and opens up vast opportunities in high-speed, robust, energy-efficient, and low-bandwidth real-time decision making.
CLIP (Contrastive Language-Image Pre-training) models developed by OpenAI have achieved outstanding results on various image recognition and retrieval tasks, displaying strong zero-shot performance. This means that they are able to perform effectively on tasks for which they have not been explicitly trained. Inspired by the success of OpenAI CLIP, a new publicly available dataset called LAION-5B was collected which resulted in the development of open ViT-H/14, ViT-G/14 models that outperform the OpenAI L/14 model. The LAION-5B dataset also released an approximate nearest neighbor index, with a web interface for search & subset creation. In this paper, we evaluate the performance of various CLIP models as zero-shot face recognizers. Our findings show that CLIP models perform well on face recognition tasks, but increasing the size of the CLIP model does not necessarily lead to improved accuracy. Additionally, we investigate the robustness of CLIP models against data poisoning attacks by testing their performance on poisoned data. Through this analysis, we aim to understand the potential consequences and misuse of search engines built using CLIP models, which could potentially function as unintentional face recognition engines.