Recent progress in scaling up large language models has shown impressive capabilities in performing few-shot learning across a wide range of text-based tasks. However, a key limitation is that these language models fundamentally lack visual perception - a crucial attribute needed to extend these models to be able to interact with the real world and solve vision tasks, such as in visual-question answering and robotics. Prior works have largely connected image to text through pretraining and/or fine-tuning on curated image-text datasets, which can be a costly and expensive process. In order to resolve this limitation, we propose a simple yet effective approach called Language-Quantized AutoEncoder (LQAE), a modification of VQ-VAE that learns to align text-image data in an unsupervised manner by leveraging pretrained language models (e.g., BERT, RoBERTa). Our main idea is to encode image as sequences of text tokens by directly quantizing image embeddings using a pretrained language codebook. We then apply random masking followed by a BERT model, and have the decoder reconstruct the original image from BERT predicted text token embeddings. By doing so, LQAE learns to represent similar images with similar clusters of text tokens, thereby aligning these two modalities without the use of aligned text-image pairs. This enables few-shot image classification with large language models (e.g., GPT-3) as well as linear classification of images based on BERT text features. To the best of our knowledge, our work is the first work that uses unaligned images for multimodal tasks by leveraging the power of pretrained language models.
相關內容
Medical image segmentation has made significant progress in recent years. Deep learning-based methods are recognized as data-hungry techniques, requiring large amounts of data with manual annotations. However, manual annotation is expensive in the field of medical image analysis, which requires domain-specific expertise. To address this challenge, few-shot learning has the potential to learn new classes from only a few examples. In this work, we propose a novel framework for few-shot medical image segmentation, termed CAT-Net, based on cross masked attention Transformer. Our proposed network mines the correlations between the support image and query image, limiting them to focus only on useful foreground information and boosting the representation capacity of both the support prototype and query features. We further design an iterative refinement framework that refines the query image segmentation iteratively and promotes the support feature in turn. We validated the proposed method on three public datasets: Abd-CT, Abd-MRI, and Card-MRI. Experimental results demonstrate the superior performance of our method compared to state-of-the-art methods and the effectiveness of each component. we will release the source codes of our method upon acceptance.
Prior works about text-to-image synthesis typically concatenated the sentence embedding with the noise vector, while the sentence embedding and the noise vector are two different factors, which control the different aspects of the generation. Simply concatenating them will entangle the latent factors and encumber the generative model. In this paper, we attempt to decompose these two factors and propose Factor Decomposed Generative Adversarial Networks~(FDGAN). To achieve this, we firstly generate images from the noise vector and then apply the sentence embedding in the normalization layer for both generator and discriminators. We also design an additive norm layer to align and fuse the text-image features. The experimental results show that decomposing the noise and the sentence embedding can disentangle latent factors in text-to-image synthesis, and make the generative model more efficient. Compared with the baseline, FDGAN can achieve better performance, while fewer parameters are used.
Recent text-to-video generation approaches rely on computationally heavy training and require large-scale video datasets. In this paper, we introduce a new task of zero-shot text-to-video generation and propose a low-cost approach (without any training or optimization) by leveraging the power of existing text-to-image synthesis methods (e.g., Stable Diffusion), making them suitable for the video domain. Our key modifications include (i) enriching the latent codes of the generated frames with motion dynamics to keep the global scene and the background time consistent; and (ii) reprogramming frame-level self-attention using a new cross-frame attention of each frame on the first frame, to preserve the context, appearance, and identity of the foreground object. Experiments show that this leads to low overhead, yet high-quality and remarkably consistent video generation. Moreover, our approach is not limited to text-to-video synthesis but is also applicable to other tasks such as conditional and content-specialized video generation, and Video Instruct-Pix2Pix, i.e., instruction-guided video editing. As experiments show, our method performs comparably or sometimes better than recent approaches, despite not being trained on additional video data. Our code will be open sourced at: //github.com/Picsart-AI-Research/Text2Video-Zero .
Sign language recognition (SLR) is a weakly supervised task that annotates sign videos as textual glosses. Recent studies show that insufficient training caused by the lack of large-scale available sign datasets becomes the main bottleneck for SLR. Most SLR works thereby adopt pretrained visual modules and develop two mainstream solutions. The multi-stream architectures extend multi-cue visual features, yielding the current SOTA performances but requiring complex designs and might introduce potential noise. Alternatively, the advanced single-cue SLR frameworks using explicit cross-modal alignment between visual and textual modalities are simple and effective, potentially competitive with the multi-cue framework. In this work, we propose a novel contrastive visual-textual transformation for SLR, CVT-SLR, to fully explore the pretrained knowledge of both the visual and language modalities. Based on the single-cue cross-modal alignment framework, we propose a variational autoencoder (VAE) for pretrained contextual knowledge while introducing the complete pretrained language module. The VAE implicitly aligns visual and textual modalities while benefiting from pretrained contextual knowledge as the traditional contextual module. Meanwhile, a contrastive cross-modal alignment algorithm is designed to explicitly enhance the consistency constraints. Extensive experiments on public datasets (PHOENIX-2014 and PHOENIX-2014T) demonstrate that our proposed CVT-SLR consistently outperforms existing single-cue methods and even outperforms SOTA multi-cue methods.
Text-to-image person retrieval aims to identify the target person based on a given textual description query. The primary challenge is to learn the mapping of visual and textual modalities into a common latent space. Prior works have attempted to address this challenge by leveraging separately pre-trained unimodal models to extract visual and textual features. However, these approaches lack the necessary underlying alignment capabilities required to match multimodal data effectively. Besides, these works use prior information to explore explicit part alignments, which may lead to the distortion of intra-modality information. To alleviate these issues, we present IRRA: a cross-modal Implicit Relation Reasoning and Aligning framework that learns relations between local visual-textual tokens and enhances global image-text matching without requiring additional prior supervision. Specifically, we first design an Implicit Relation Reasoning module in a masked language modeling paradigm. This achieves cross-modal interaction by integrating the visual cues into the textual tokens with a cross-modal multimodal interaction encoder. Secondly, to globally align the visual and textual embeddings, Similarity Distribution Matching is proposed to minimize the KL divergence between image-text similarity distributions and the normalized label matching distributions. The proposed method achieves new state-of-the-art results on all three public datasets, with a notable margin of about 3%-9% for Rank-1 accuracy compared to prior methods.
The imsets of \citet{studeny2006probabilistic} are an algebraic method for representing conditional independence models. They have many attractive properties when applied to such models, and they are particularly nice for working with directed acyclic graph (DAG) models. In particular, the `standard' imset for a DAG is in one-to-one correspondence with the independences it induces, and hence is a label for its Markov equivalence class. We first present a proposed extension to standard imsets for maximal ancestral graph (MAG) models, using the parameterizing set representation of \citet{hu2020faster}. We show that for many such graphs our proposed imset is \emph{perfectly Markovian} with respect to the graph, including \emph{simple} MAGs, as well as for a large class of purely bidirected models. Thus providing a scoring criteria by measuring the discrepancy for a list of independences that define the model; this gives an alternative to the usual BIC score that is much easier to compute. We also show that, of independence models that do represent the MAG, the one we give is the simplest possible, in a manner we make precise. Unfortunately, for some graphs the representation does not represent all the independences in the model, and in certain cases does not represent any at all. For these general MAGs, we refine the reduced ordered local Markov property \citep{richardlocalmarkov} by a novel graphical tool called \emph{power DAGs}, and this results in an imset that induces the correct model and which, under a mild condition, can be constructed in polynomial time.
With the rise of powerful pre-trained vision-language models like CLIP, it becomes essential to investigate ways to adapt these models to downstream datasets. A recently proposed method named Context Optimization (CoOp) introduces the concept of prompt learning -- a recent trend in NLP -- to the vision domain for adapting pre-trained vision-language models. Specifically, CoOp turns context words in a prompt into a set of learnable vectors and, with only a few labeled images for learning, can achieve huge improvements over intensively-tuned manual prompts. In our study we identify a critical problem of CoOp: the learned context is not generalizable to wider unseen classes within the same dataset, suggesting that CoOp overfits base classes observed during training. To address the problem, we propose Conditional Context Optimization (CoCoOp), which extends CoOp by further learning a lightweight neural network to generate for each image an input-conditional token (vector). Compared to CoOp's static prompts, our dynamic prompts adapt to each instance and are thus less sensitive to class shift. Extensive experiments show that CoCoOp generalizes much better than CoOp to unseen classes, even showing promising transferability beyond a single dataset; and yields stronger domain generalization performance as well. Code is available at //github.com/KaiyangZhou/CoOp.
The notion of "in-domain data" in NLP is often over-simplistic and vague, as textual data varies in many nuanced linguistic aspects such as topic, style or level of formality. In addition, domain labels are many times unavailable, making it challenging to build domain-specific systems. We show that massive pre-trained language models implicitly learn sentence representations that cluster by domains without supervision -- suggesting a simple data-driven definition of domains in textual data. We harness this property and propose domain data selection methods based on such models, which require only a small set of in-domain monolingual data. We evaluate our data selection methods for neural machine translation across five diverse domains, where they outperform an established approach as measured by both BLEU and by precision and recall of sentence selection with respect to an oracle.
Few-shot image classification aims to classify unseen classes with limited labeled samples. Recent works benefit from the meta-learning process with episodic tasks and can fast adapt to class from training to testing. Due to the limited number of samples for each task, the initial embedding network for meta learning becomes an essential component and can largely affects the performance in practice. To this end, many pre-trained methods have been proposed, and most of them are trained in supervised way with limited transfer ability for unseen classes. In this paper, we proposed to train a more generalized embedding network with self-supervised learning (SSL) which can provide slow and robust representation for downstream tasks by learning from the data itself. We evaluate our work by extensive comparisons with previous baseline methods on two few-shot classification datasets ({\em i.e.,} MiniImageNet and CUB). Based on the evaluation results, the proposed method achieves significantly better performance, i.e., improve 1-shot and 5-shot tasks by nearly \textbf{3\%} and \textbf{4\%} on MiniImageNet, by nearly \textbf{9\%} and \textbf{3\%} on CUB. Moreover, the proposed method can gain the improvement of (\textbf{15\%}, \textbf{13\%}) on MiniImageNet and (\textbf{15\%}, \textbf{8\%}) on CUB by pretraining using more unlabeled data. Our code will be available at \hyperref[//github.com/phecy/SSL-FEW-SHOT.]{//github.com/phecy/ssl-few-shot.}
Joint image-text embedding is the bedrock for most Vision-and-Language (V+L) tasks, where multimodality inputs are jointly processed for visual and textual understanding. In this paper, we introduce UNITER, a UNiversal Image-TExt Representation, learned through large-scale pre-training over four image-text datasets (COCO, Visual Genome, Conceptual Captions, and SBU Captions), which can power heterogeneous downstream V+L tasks with joint multimodal embeddings. We design three pre-training tasks: Masked Language Modeling (MLM), Image-Text Matching (ITM), and Masked Region Modeling (MRM, with three variants). Different from concurrent work on multimodal pre-training that apply joint random masking to both modalities, we use conditioned masking on pre-training tasks (i.e., masked language/region modeling is conditioned on full observation of image/text). Comprehensive analysis shows that conditioned masking yields better performance than unconditioned masking. We also conduct a thorough ablation study to find an optimal setting for the combination of pre-training tasks. Extensive experiments show that UNITER achieves new state of the art across six V+L tasks (over nine datasets), including Visual Question Answering, Image-Text Retrieval, Referring Expression Comprehension, Visual Commonsense Reasoning, Visual Entailment, and NLVR2.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191