The rapid advancement of large language models (LLMs) has accelerated the emergence of in-context learning (ICL) as a cutting-edge approach in the natural language processing domain. Recently, ICL has been employed in visual understanding tasks, such as semantic segmentation and image captioning, yielding promising results. However, existing visual ICL framework can not enable producing content across multiple modalities, which limits their potential usage scenarios. To address this issue, we present a new ICL framework for visual understanding with multi-modal output enabled. First, we quantize and embed both text and visual prompt into a unified representational space, structured as interleaved in-context sequences. Then a decoder-only sparse transformer architecture is employed to perform generative modeling on them, facilitating in-context learning. Thanks to this design, the model is capable of handling in-context vision understanding tasks with multimodal output in a unified pipeline. Experimental results demonstrate that our model achieves competitive performance compared with specialized models and previous ICL baselines. Overall, our research takes a further step toward unified multimodal in-context learning.
The MineRL BASALT competition has served to catalyze advances in learning from human feedback through four hard-to-specify tasks in Minecraft, such as create and photograph a waterfall. Given the completion of two years of BASALT competitions, we offer to the community a formalized benchmark through the BASALT Evaluation and Demonstrations Dataset (BEDD), which serves as a resource for algorithm development and performance assessment. BEDD consists of a collection of 26 million image-action pairs from nearly 14,000 videos of human players completing the BASALT tasks in Minecraft. It also includes over 3,000 dense pairwise human evaluations of human and algorithmic agents. These comparisons serve as a fixed, preliminary leaderboard for evaluating newly-developed algorithms. To enable this comparison, we present a streamlined codebase for benchmarking new algorithms against the leaderboard. In addition to presenting these datasets, we conduct a detailed analysis of the data from both datasets to guide algorithm development and evaluation. The released code and data are available at https://github.com/minerllabs/basalt-benchmark .
We present a diffusion-based video editing framework, namely DiffusionAtlas, which can achieve both frame consistency and high fidelity in editing video object appearance. Despite the success in image editing, diffusion models still encounter significant hindrances when it comes to video editing due to the challenge of maintaining spatiotemporal consistency in the object's appearance across frames. On the other hand, atlas-based techniques allow propagating edits on the layered representations consistently back to frames. However, they often struggle to create editing effects that adhere correctly to the user-provided textual or visual conditions due to the limitation of editing the texture atlas on a fixed UV mapping field. Our method leverages a visual-textual diffusion model to edit objects directly on the diffusion atlases, ensuring coherent object identity across frames. We design a loss term with atlas-based constraints and build a pretrained text-driven diffusion model as pixel-wise guidance for refining shape distortions and correcting texture deviations. Qualitative and quantitative experiments show that our method outperforms state-of-the-art methods in achieving consistent high-fidelity video-object editing.
Breast ultrasound (BUS) is a critical diagnostic tool in the field of breast imaging, aiding in the early detection and characterization of breast abnormalities. Interpreting breast ultrasound images commonly involves creating comprehensive medical reports, containing vital information to promptly assess the patient's condition. However, the ultrasound imaging system necessitates capturing multiple images of various parts to compile a single report, presenting a time-consuming challenge. To address this problem, we propose the integration of multiple image analysis tools through a LangChain using Large Language Models (LLM), into the breast reporting process. Through a combination of designated tools and text generation through LangChain, our method can accurately extract relevant features from ultrasound images, interpret them in a clinical context, and produce comprehensive and standardized reports. This approach not only reduces the burden on radiologists and healthcare professionals but also enhances the consistency and quality of reports. The extensive experiments shows that each tools involved in the proposed method can offer qualitatively and quantitatively significant results. Furthermore, clinical evaluation on the generated reports demonstrates that the proposed method can make report in clinically meaningful way.
General purpose segmentation models are able to generate (semantic) segmentation masks from a variety of prompts, including visual (points, boxed, etc.) and textual (object names) ones. In particular, input images are pre-processed by an image encoder to obtain embedding vectors which are later used for mask predictions. Existing adversarial attacks target the end-to-end tasks, i.e. aim at altering the segmentation mask predicted for a specific image-prompt pair. However, this requires running an individual attack for each new prompt for the same image. We propose instead to generate prompt-agnostic adversarial attacks by maximizing the $\ell_2$-distance, in the latent space, between the embedding of the original and perturbed images. Since the encoding process only depends on the image, distorted image representations will cause perturbations in the segmentation masks for a variety of prompts. We show that even imperceptible $\ell_\infty$-bounded perturbations of radius $\epsilon=1/255$ are often sufficient to drastically modify the masks predicted with point, box and text prompts by recently proposed foundation models for segmentation. Moreover, we explore the possibility of creating universal, i.e. non image-specific, attacks which can be readily applied to any input without further computational cost.
In this paper, we study a challenging task of zero-shot referring image segmentation. This task aims to identify the instance mask that is most related to a referring expression without training on pixel-level annotations. Previous research takes advantage of pre-trained cross-modal models, e.g., CLIP, to align instance-level masks with referring expressions. %Yet, CLIP only considers image-text pair level alignment, which neglects fine-grained image region and complex sentence matching. Yet, CLIP only considers the global-level alignment of image-text pairs, neglecting fine-grained matching between the referring sentence and local image regions. To address this challenge, we introduce a Text Augmented Spatial-aware (TAS) zero-shot referring image segmentation framework that is training-free and robust to various visual encoders. TAS incorporates a mask proposal network for instance-level mask extraction, a text-augmented visual-text matching score for mining the image-text correlation, and a spatial rectifier for mask post-processing. Notably, the text-augmented visual-text matching score leverages a $P$ score and an $N$-score in addition to the typical visual-text matching score. The $P$-score is utilized to close the visual-text domain gap through a surrogate captioning model, where the score is computed between the surrogate model-generated texts and the referring expression. The $N$-score considers the fine-grained alignment of region-text pairs via negative phrase mining, encouraging the masked image to be repelled from the mined distracting phrases. Extensive experiments are conducted on various datasets, including RefCOCO, RefCOCO+, and RefCOCOg. The proposed method clearly outperforms state-of-the-art zero-shot referring image segmentation methods.
Retrieving spatial information and understanding the semantic information of the surroundings are important for Bird's-Eye-View (BEV) semantic segmentation. In the application of autonomous driving, autonomous vehicles need to be aware of their surroundings to drive safely. However, current BEV semantic segmentation techniques, deep Convolutional Neural Networks (CNNs) and transformers, have difficulties in obtaining the global semantic relationships of the surroundings at the early layers of the network. In this paper, we propose to incorporate a novel Residual Graph Convolutional (RGC) module in deep CNNs to acquire both the global information and the region-level semantic relationship in the multi-view image domain. Specifically, the RGC module employs a non-overlapping graph space projection to efficiently project the complete BEV information into graph space. It then builds interconnected spatial and channel graphs to extract spatial information between each node and channel information within each node (i.e., extract contextual relationships of the global features). Furthermore, it uses a downsample residual process to enhance the coordinate feature reuse to maintain the global information. The segmentation data augmentation and alignment module helps to simultaneously augment and align BEV features and ground truth to geometrically preserve their alignment to achieve better segmentation results. Our experimental results on the nuScenes benchmark dataset demonstrate that the RGC network outperforms four state-of-the-art networks and its four variants in terms of IoU and mIoU. The proposed RGC network achieves a higher mIoU of 3.1% than the best state-of-the-art network, BEVFusion. Code and models will be released.
Large multimodal models demonstrate remarkable generalist ability to perform diverse multimodal tasks in a zero-shot manner. Large-scale web-based image-text pairs contribute fundamentally to this success, but suffer from excessive noise. Recent studies use alternative captions synthesized by captioning models and have achieved notable benchmark performance. However, our experiments reveal significant Scalability Deficiency and World Knowledge Loss issues in models trained with synthetic captions, which have been largely obscured by their initial benchmark success. Upon closer examination, we identify the root cause as the overly-simplified language structure and lack of knowledge details in existing synthetic captions. To provide higher-quality and more scalable multimodal pretraining data, we propose CapsFusion, an advanced framework that leverages large language models to consolidate and refine information from both web-based image-text pairs and synthetic captions. Extensive experiments show that CapsFusion captions exhibit remarkable all-round superiority over existing captions in terms of model performance (e.g., 18.8 and 18.3 improvements in CIDEr score on COCO and NoCaps), sample efficiency (requiring 11-16 times less computation than baselines), world knowledge depth, and scalability. These effectiveness, efficiency and scalability advantages position CapsFusion as a promising candidate for future scaling of LMM training.
Full-spectrum out-of-distribution (F-OOD) detection aims to accurately recognize in-distribution (ID) samples while encountering semantic and covariate shifts simultaneously. However, existing out-of-distribution (OOD) detectors tend to overfit the covariance information and ignore intrinsic semantic correlation, inadequate for adapting to complex domain transformations. To address this issue, we propose a Likelihood-Aware Semantic Alignment (LSA) framework to promote the image-text correspondence into semantically high-likelihood regions. LSA consists of an offline Gaussian sampling strategy which efficiently samples semantic-relevant visual embeddings from the class-conditional Gaussian distribution, and a bidirectional prompt customization mechanism that adjusts both ID-related and negative context for discriminative ID/OOD boundary. Extensive experiments demonstrate the remarkable OOD detection performance of our proposed LSA especially on the intractable Near-OOD setting, surpassing existing methods by a margin of $15.26\%$ and $18.88\%$ on two F-OOD benchmarks, respectively.
The choice of input text prompt plays a critical role in the performance of Vision-Language Pretrained (VLP) models such as CLIP. We present APoLLo, a unified multi-modal approach that combines Adapter and Prompt learning for Vision-Language models. Our method is designed to substantially improve the generalization capabilities of VLP models when they are fine-tuned in a few-shot setting. We introduce trainable cross-attention-based adapter layers in conjunction with vision and language encoders to strengthen the alignment between the two modalities. We enforce consistency between the respective encoder branches (receiving augmented inputs) to prevent overfitting in downstream tasks. Our method is evaluated on three representative tasks: generalization to novel classes, cross-dataset evaluation, and unseen domain shifts. In practice, APoLLo achieves a relative gain up to 6.03% over MaPLe (SOTA) on novel classes for 10 diverse image recognition datasets.