Recently, groundbreaking results have been presented on open-vocabulary semantic image segmentation. Such methods segment each pixel in an image into arbitrary categories provided at run-time in the form of text prompts, as opposed to a fixed set of classes defined at training time. In this work, we present a zero-shot volumetric open-vocabulary semantic scene segmentation method. Our method builds on the insight that we can fuse image features from a vision-language model into a neural implicit representation. We show that the resulting feature field can be segmented into different classes by assigning points to natural language text prompts. The implicit volumetric representation enables us to segment the scene both in 3D and 2D by rendering feature maps from any given viewpoint of the scene. We show that our method works on noisy real-world data and can run in real-time on live sensor data dynamically adjusting to text prompts. We also present quantitative comparisons on the ScanNet dataset.
Most existing scene text detectors focus on detecting characters or words that only capture partial text messages due to missing contextual information. For a better understanding of text in scenes, it is more desired to detect contextual text blocks (CTBs) which consist of one or multiple integral text units (e.g., characters, words, or phrases) in natural reading order and transmit certain complete text messages. This paper presents contextual text detection, a new setup that detects CTBs for better understanding of texts in scenes. We formulate the new setup by a dual detection task which first detects integral text units and then groups them into a CTB. To this end, we design a novel scene text clustering technique that treats integral text units as tokens and groups them (belonging to the same CTB) into an ordered token sequence. In addition, we create two datasets SCUT-CTW-Context and ReCTS-Context to facilitate future research, where each CTB is well annotated by an ordered sequence of integral text units. Further, we introduce three metrics that measure contextual text detection in local accuracy, continuity, and global accuracy. Extensive experiments show that our method accurately detects CTBs which effectively facilitates downstream tasks such as text classification and translation. The project is available at https://sg-vilab.github.io/publication/xue2022contextual/.
Text-based person search (TBPS) is of significant importance in intelligent surveillance, which aims to retrieve pedestrian images with high semantic relevance to a given text description. This retrieval task is characterized with both modal heterogeneity and fine-grained matching. To implement this task, one needs to extract multi-scale features from both image and text domains, and then perform the cross-modal alignment. However, most existing approaches only consider the alignment confined at their individual scales, e.g., an image-sentence or a region-phrase scale. Such a strategy adopts the presumable alignment in feature extraction, while overlooking the cross-scale alignment, e.g., image-phrase. In this paper, we present a transformer-based model to extract multi-scale representations, and perform Asymmetric Cross-Scale Alignment (ACSA) to precisely align the two modalities. Specifically, ACSA consists of a global-level alignment module and an asymmetric cross-attention module, where the former aligns an image and texts on a global scale, and the latter applies the cross-attention mechanism to dynamically align the cross-modal entities in region/image-phrase scales. Extensive experiments on two benchmark datasets CUHK-PEDES and RSTPReid demonstrate the effectiveness of our approach. Codes are available at \href{url}{https://github.com/mul-hjh/ACSA}.
Pre-trained vision-language models like CLIP have recently shown superior performances on various downstream tasks, including image classification and segmentation. However, in fine-grained image re-identification (ReID), the labels are indexes, lacking concrete text descriptions. Therefore, it remains to be determined how such models could be applied to these tasks. This paper first finds out that simply fine-tuning the visual model initialized by the image encoder in CLIP, has already obtained competitive performances in various ReID tasks. Then we propose a two-stage strategy to facilitate a better visual representation. The key idea is to fully exploit the cross-modal description ability in CLIP through a set of learnable text tokens for each ID and give them to the text encoder to form ambiguous descriptions. In the first training stage, image and text encoders from CLIP keep fixed, and only the text tokens are optimized from scratch by the contrastive loss computed within a batch. In the second stage, the ID-specific text tokens and their encoder become static, providing constraints for fine-tuning the image encoder. With the help of the designed loss in the downstream task, the image encoder is able to represent data as vectors in the feature embedding accurately. The effectiveness of the proposed strategy is validated on several datasets for the person or vehicle ReID tasks. Code is available at https://github.com/Syliz517/CLIP-ReID.
Text classification methods have been widely investigated as a way to detect content of low credibility: fake news, social media bots, propaganda, etc. Quite accurate models (likely based on deep neural networks) help in moderating public electronic platforms and often cause content creators to face rejection of their submissions or removal of already published texts. Having the incentive to evade further detection, content creators try to come up with a slightly modified version of the text (known as an attack with an adversarial example) that exploit the weaknesses of classifiers and result in a different output. Here we introduce BODEGA: a benchmark for testing both victim models and attack methods on four misinformation detection tasks in an evaluation framework designed to simulate real use-cases of content moderation. We also systematically test the robustness of popular text classifiers against available attacking techniques and discover that, indeed, in some cases barely significant changes in input text can mislead the models. We openly share the BODEGA code and data in hope of enhancing the comparability and replicability of further research in this area.
Leveraging medical record information in the era of big data and machine learning comes with the caveat that data must be cleaned and deidentified. Facilitating data sharing and harmonization for multi-center collaborations are particularly difficult when protected health information (PHI) is contained or embedded in image meta-data. We propose a novel library in the Python framework, called PyLogik, to help alleviate this issue for ultrasound images, which are particularly challenging because of the frequent inclusion of PHI directly on the images. PyLogik processes the image volumes through a series of text detection/extraction, filtering, thresholding, morphological and contour comparisons. This methodology deidentifies the images, reduces file sizes, and prepares image volumes for applications in deep learning and data sharing. To evaluate its effectiveness in the identification of regions of interest (ROI), a random sample of 50 cardiac ultrasounds (echocardiograms) were processed through PyLogik, and the outputs were compared with the manual segmentations by an expert user. The Dice coefficient of the two approaches achieved an average value of 0.976. Next, an investigation was conducted to ascertain the degree of information compression achieved using the algorithm. Resultant data was found to be on average approximately 72% smaller after processing by PyLogik. Our results suggest that PyLogik is a viable methodology for ultrasound data cleaning and deidentification, determining ROI, and file compression which will facilitate efficient storage, use, and dissemination of ultrasound data.
Recent breakthroughs in Vision-Language (V&L) joint research have achieved remarkable results in various text-driven tasks. High-quality Text-to-video (T2V), a task that has been long considered mission-impossible, was proven feasible with reasonably good results in latest works. However, the resulting videos often have undesired artifacts largely because the system is purely data-driven and agnostic to the physical laws. To tackle this issue and further push T2V towards high-level physical realism, we present an autonomous data generation technique and a dataset, which intend to narrow the gap with a large number of multi-modal, 3D Text-to-Video/Simulation (T2V/S) data. In the dataset, we provide high-resolution 3D physical simulations for both solids and fluids, along with textual descriptions of the physical phenomena. We take advantage of state-of-the-art physical simulation methods (i) Incremental Potential Contact (IPC) and (ii) Material Point Method (MPM) to simulate diverse scenarios, including elastic deformations, material fractures, collisions, turbulence, etc. Additionally, high-quality, multi-view rendering videos are supplied for the benefit of T2V, Neural Radiance Fields (NeRF), and other communities. This work is the first step towards fully automated Text-to-Video/Simulation (T2V/S). Live examples and subsequent work are at https://sites.google.com/view/tpa-net.
When annotators label data, a key metric for quality assurance is inter-annotator agreement (IAA): the extent to which annotators agree on their labels. Though many IAA measures exist for simple categorical and ordinal labeling tasks, relatively little work has considered more complex labeling tasks, such as structured, multi-object, and free-text annotations. Krippendorff's alpha, best known for use with simpler labeling tasks, does have a distance-based formulation with broader applicability, but little work has studied its efficacy and consistency across complex annotation tasks. We investigate the design and evaluation of IAA measures for complex annotation tasks, with evaluation spanning seven diverse tasks: image bounding boxes, image keypoints, text sequence tagging, ranked lists, free text translations, numeric vectors, and syntax trees. We identify the difficulty of interpretability and the complexity of choosing a distance function as key obstacles in applying Krippendorff's alpha generally across these tasks. We propose two novel, more interpretable measures, showing they yield more consistent IAA measures across tasks and annotation distance functions.
Labeling mistakes are frequently encountered in real-world applications. If not treated well, the labeling mistakes can deteriorate the classification performances of a model seriously. To address this issue, we propose an improved Naive Bayes method for text classification. It is analytically simple and free of subjective judgements on the correct and incorrect labels. By specifying the generating mechanism of incorrect labels, we optimize the corresponding log-likelihood function iteratively by using an EM algorithm. Our simulation and experiment results show that the improved Naive Bayes method greatly improves the performances of the Naive Bayes method with mislabeled data.
Typical layout-to-image synthesis (LIS) models generate images for a closed set of semantic classes, e.g., 182 common objects in COCO-Stuff. In this work, we explore the freestyle capability of the model, i.e., how far can it generate unseen semantics (e.g., classes, attributes, and styles) onto a given layout, and call the task Freestyle LIS (FLIS). Thanks to the development of large-scale pre-trained language-image models, a number of discriminative models (e.g., image classification and object detection) trained on limited base classes are empowered with the ability of unseen class prediction. Inspired by this, we opt to leverage large-scale pre-trained text-to-image diffusion models to achieve the generation of unseen semantics. The key challenge of FLIS is how to enable the diffusion model to synthesize images from a specific layout which very likely violates its pre-learned knowledge, e.g., the model never sees "a unicorn sitting on a bench" during its pre-training. To this end, we introduce a new module called Rectified Cross-Attention (RCA) that can be conveniently plugged in the diffusion model to integrate semantic masks. This "plug-in" is applied in each cross-attention layer of the model to rectify the attention maps between image and text tokens. The key idea of RCA is to enforce each text token to act on the pixels in a specified region, allowing us to freely put a wide variety of semantics from pre-trained knowledge (which is general) onto the given layout (which is specific). Extensive experiments show that the proposed diffusion network produces realistic and freestyle layout-to-image generation results with diverse text inputs, which has a high potential to spawn a bunch of interesting applications. Code is available at https://github.com/essunny310/FreestyleNet.