As the capabilities of large multimodal models (LMMs) continue to advance, evaluating the performance of LMMs emerges as an increasing need. Additionally, there is an even larger gap in evaluating the advanced knowledge and reasoning abilities of LMMs in non-English contexts such as Chinese. We introduce CMMMU, a new Chinese Massive Multi-discipline Multimodal Understanding benchmark designed to evaluate LMMs on tasks demanding college-level subject knowledge and deliberate reasoning in a Chinese context. CMMMU is inspired by and strictly follows the annotation and analysis pattern of MMMU. CMMMU includes 12k manually collected multimodal questions from college exams, quizzes, and textbooks, covering six core disciplines: Art & Design, Business, Science, Health & Medicine, Humanities & Social Science, and Tech & Engineering, like its companion, MMMU. These questions span 30 subjects and comprise 39 highly heterogeneous image types, such as charts, diagrams, maps, tables, music sheets, and chemical structures. CMMMU focuses on complex perception and reasoning with domain-specific knowledge in the Chinese context. We evaluate 11 open-source LLMs and one proprietary GPT-4V(ision). Even GPT-4V only achieves accuracies of 42%, indicating a large space for improvement. CMMMU will boost the community to build the next-generation LMMs towards expert artificial intelligence and promote the democratization of LMMs by providing diverse language contexts.
Skull base surgery is a demanding field in which surgeons operate in and around the skull while avoiding critical anatomical structures including nerves and vasculature. While image-guided surgical navigation is the prevailing standard, limitation still exists requiring personalized planning and recognizing the irreplaceable role of a skilled surgeon. This paper presents a collaboratively controlled robotic system tailored for assisted drilling in skull base surgery. Our central hypothesis posits that this collaborative system, enriched with haptic assistive modes to enforce virtual fixtures, holds the potential to significantly enhance surgical safety, streamline efficiency, and alleviate the physical demands on the surgeon. The paper describes the intricate system development work required to enable these virtual fixtures through haptic assistive modes. To validate our system's performance and effectiveness, we conducted initial feasibility experiments involving a medical student and two experienced surgeons. The experiment focused on drilling around critical structures following cortical mastoidectomy, utilizing dental stone phantom and cadaveric models. Our experimental results demonstrate that our proposed haptic feedback mechanism enhances the safety of drilling around critical structures compared to systems lacking haptic assistance. With the aid of our system, surgeons were able to safely skeletonize the critical structures without breaching any critical structure even under obstructed view of the surgical site.
Accurate cloud recognition and warning are crucial for various applications, including in-flight support, weather forecasting, and climate research. However, recent deep learning algorithms have predominantly focused on detecting cloud regions in satellite imagery, with insufficient attention to the specificity required for accurate cloud recognition. This limitation inspired us to develop the novel FY-4A-Himawari-8 (FYH) dataset, which includes nine distinct cloud categories and uses precise domain adaptation methods to align 70,419 image-label pairs in terms of projection, temporal resolution, and spatial resolution, thereby facilitating the training of supervised deep learning networks. Given the complexity and diversity of cloud formations, we have thoroughly analyzed the challenges inherent to cloud recognition tasks, examining the intricate characteristics and distribution of the data. To effectively address these challenges, we designed a Distribution-aware Interactive-Attention Network (DIAnet), which preserves pixel-level details through a high-resolution branch and a parallel multi-resolution cross-branch. We also integrated a distribution-aware loss (DAL) to mitigate the imbalance across cloud categories. An Interactive Attention Module (IAM) further enhances the robustness of feature extraction combined with spatial and channel information. Empirical evaluations on the FYH dataset demonstrate that our method outperforms other cloud recognition networks, achieving superior performance in terms of mean Intersection over Union (mIoU). The code for implementing DIAnet is available at https://github.com/icey-zhang/DIAnet.
While deep neural networks have achieved remarkable performance, data augmentation has emerged as a crucial strategy to mitigate overfitting and enhance network performance. These techniques hold particular significance in industrial manufacturing contexts. Recently, image mixing-based methods have been introduced, exhibiting improved performance on public benchmark datasets. However, their application to industrial tasks remains challenging. The manufacturing environment generates massive amounts of unlabeled data on a daily basis, with only a few instances of abnormal data occurrences. This leads to severe data imbalance. Thus, creating well-balanced datasets is not straightforward due to the high costs associated with labeling. Nonetheless, this is a crucial step for enhancing productivity. For this reason, we introduce ContextMix, a method tailored for industrial applications and benchmark datasets. ContextMix generates novel data by resizing entire images and integrating them into other images within the batch. This approach enables our method to learn discriminative features based on varying sizes from resized images and train informative secondary features for object recognition using occluded images. With the minimal additional computation cost of image resizing, ContextMix enhances performance compared to existing augmentation techniques. We evaluate its effectiveness across classification, detection, and segmentation tasks using various network architectures on public benchmark datasets. Our proposed method demonstrates improved results across a range of robustness tasks. Its efficacy in real industrial environments is particularly noteworthy, as demonstrated using the passive component dataset.
Weakly supervised text-based person re-identification (TPRe-ID) seeks to retrieve images of a target person using textual descriptions, without relying on identity annotations and is more challenging and practical. The primary challenge is the intra-class differences, encompassing intra-modal feature variations and cross-modal semantic gaps. Prior works have focused on instance-level samples and ignored prototypical features of each person which are intrinsic and invariant. Toward this, we propose a Cross-Modal Prototypical Contrastive Learning (CPCL) method. In practice, the CPCL introduces the CLIP model to weakly supervised TPRe-ID for the first time, mapping visual and textual instances into a shared latent space. Subsequently, the proposed Prototypical Multi-modal Memory (PMM) module captures associations between heterogeneous modalities of image-text pairs belonging to the same person through the Hybrid Cross-modal Matching (HCM) module in a many-to-many mapping fashion. Moreover, the Outlier Pseudo Label Mining (OPLM) module further distinguishes valuable outlier samples from each modality, enhancing the creation of more reliable clusters by mining implicit relationships between image-text pairs. Experimental results demonstrate that our proposed CPCL attains state-of-the-art performance on all three public datasets, with a significant improvement of 11.58%, 8.77% and 5.25% in Rank@1 accuracy on CUHK-PEDES, ICFG-PEDES and RSTPReid datasets, respectively. The code is available at https://github.com/codeGallery24/CPCL.
A standard treatment protocol for breast cancer entails administering neoadjuvant therapy followed by surgical removal of the tumor and surrounding tissue. Pathologists typically rely on cabinet X-ray radiographs, known as Faxitron, to examine the excised breast tissue and diagnose the extent of residual disease. However, accurately determining the location, size, and focality of residual cancer can be challenging, and incorrect assessments can lead to clinical consequences. The utilization of automated methods can improve the histopathology process, allowing pathologists to choose regions for sampling more effectively and precisely. Despite the recognized necessity, there are currently no such methods available. Training such automated detection models require accurate ground truth labels on ex-vivo radiology images, which can be acquired through registering Faxitron and histopathology images and mapping the extent of cancer from histopathology to x-ray images. This study introduces a deep learning-based image registration approach trained on mono-modal synthetic image pairs. The models were trained using data from 50 women who received neoadjuvant chemotherapy and underwent surgery. The results demonstrate that our method is faster and yields significantly lower average landmark error ($2.1\pm1.96$ mm) over the state-of-the-art iterative ($4.43\pm4.1$ mm) and deep learning ($4.02\pm3.15$ mm) approaches. Improved performance of our approach in integrating radiology and pathology information facilitates generating large datasets, which allows training models for more accurate breast cancer detection.
Advancements in medical imaging and endovascular grafting have facilitated minimally invasive treatments for aortic diseases. Accurate 3D segmentation of the aorta and its branches is crucial for interventions, as inaccurate segmentation can lead to erroneous surgical planning and endograft construction. Previous methods simplified aortic segmentation as a binary image segmentation problem, overlooking the necessity of distinguishing between individual aortic branches. In this paper, we introduce Context Infused Swin-UNet (CIS-UNet), a deep learning model designed for multi-class segmentation of the aorta and thirteen aortic branches. Combining the strengths of Convolutional Neural Networks (CNNs) and Swin transformers, CIS-UNet adopts a hierarchical encoder-decoder structure comprising a CNN encoder, symmetric decoder, skip connections, and a novel Context-aware Shifted Window Self-Attention (CSW-SA) as the bottleneck block. Notably, CSW-SA introduces a unique utilization of the patch merging layer, distinct from conventional Swin transformers. It efficiently condenses the feature map, providing a global spatial context and enhancing performance when applied at the bottleneck layer, offering superior computational efficiency and segmentation accuracy compared to the Swin transformers. We trained our model on computed tomography (CT) scans from 44 patients and tested it on 15 patients. CIS-UNet outperformed the state-of-the-art SwinUNetR segmentation model, which is solely based on Swin transformers, by achieving a superior mean Dice coefficient of 0.713 compared to 0.697, and a mean surface distance of 2.78 mm compared to 3.39 mm. CIS-UNet's superior 3D aortic segmentation offers improved precision and optimization for planning endovascular treatments. Our dataset and code will be publicly available.
Purpose: To evaluate the quality of deep learning reconstruction for prospectively accelerated intraoperative magnetic resonance imaging (iMRI) during resective brain tumor surgery. Materials and Methods: Accelerated iMRI was performed during brain surgery using dual surface coils positioned around the area of resection. A deep learning (DL) model was trained on the fastMRI neuro dataset to mimic the data from the iMRI protocol. Evaluation was performed on imaging material from 40 patients imaged between 01.11.2021 - 01.06.2023 that underwent iMRI during tumor resection surgery. A comparative analysis was conducted between the conventional compressed sense (CS) method and the trained DL reconstruction method. Blinded evaluation of multiple image quality metrics was performed by two working neuro-radiologists and a working neurosurgeon on a 1 to 5 Likert scale (1=non diagnostic, 2=poor, 3=acceptable, 4=good, 5=excellent), and the favored reconstruction variant. Results: The DL reconstruction was strongly favored or favored over the CS reconstruction for 33/40, 39/40, and 8/40 of cases for reader 1, 2, and 3, respectively. Two of three readers consistently assigned higher ratings for the DL reconstructions, and the DL reconstructions had a higher score than their respective CS counterparts for 72%, 72%, and 14% of the cases for reader 1, 2, and 3, respectively. Still, the DL reconstructions exhibited shortcomings such as a striping artifact and reduced signal. Conclusion: DL shows promise to allow for high-quality reconstructions of intraoperative MRI with equal to or improved perceived spatial resolution, signal-to-noise ratio, diagnostic confidence, diagnostic conspicuity, and spatial resolution compared to compressed sense.
Latent representations are used extensively for downstream tasks, such as visualization, interpolation or feature extraction of deep learning models. Invariant and equivariant neural networks are powerful and well-established models for enforcing inductive biases. In this paper, we demonstrate that the inductive bias imposed on the by an equivariant model must also be taken into account when using latent representations. We show how not accounting for the inductive biases leads to decreased performance on downstream tasks, and vice versa, how accounting for inductive biases can be done effectively by using an invariant projection of the latent representations. We propose principles for how to choose such a projection, and show the impact of using these principles in two common examples: First, we study a permutation equivariant variational auto-encoder trained for molecule graph generation; here we show that invariant projections can be designed that incur no loss of information in the resulting invariant representation. Next, we study a rotation-equivariant representation used for image classification. Here, we illustrate how random invariant projections can be used to obtain an invariant representation with a high degree of retained information. In both cases, the analysis of invariant latent representations proves superior to their equivariant counterparts. Finally, we illustrate that the phenomena documented here for equivariant neural networks have counterparts in standard neural networks where invariance is encouraged via augmentation. Thus, while these ambiguities may be known by experienced developers of equivariant models, we make both the knowledge as well as effective tools to handle the ambiguities available to the broader community.
Vision-language models (VLMs) excel in zero-shot recognition but exhibit drastically imbalanced performance across visual concepts. For example, CLIP, despite an impressive mean zero-shot accuracy on ImageNet (72.7%), yields $<$10% on ten concepts (e.g., gyromitra and night snake), presumably, because these concepts are under-represented in VLMs' imbalanced pretraining data. Yet, assessing this imbalance is challenging as it is non-trivial to calculate the frequency of specific concepts within VLMs' large-scale pretraining data. Our work makes the first attempt to measure the concept frequency by analyzing pretraining texts. We use off-the-shelf language models to help count relevant texts that contain synonyms of the given concepts and resolve linguistic ambiguity. We confirm that popular VLM datasets like LAION indeed exhibit long-tailed concept distributions, which strongly correlate with per-class accuracies. Further, contemporary multimodal systems, e.g., visual chatbots and text-to-image generators, also struggle with the rare concepts identified by our method. To mitigate VLMs' imbalanced performance in zero-shot recognition, we propose REtrieval-Augmented Learning REAL. First, instead of prompting VLMs using the original class names, REAL uses their most frequent synonyms found in VLMs' pretraining texts. This already outperforms human-engineered and LLM-generated prompts over nine benchmark datasets, likely because VLMs have seen more images associated with the frequently used synonyms. Second, REAL uses all the concept synonyms to retrieve a small, class-balanced set of pretraining data to train a robust classifier. REAL surpasses the recent retrieval-augmented solution REACT, using 400x less storage and 10,000x less training time!