Large-scale pre-trained models have demonstrated impressive performance in vision and language tasks within open-world scenarios. Due to the lack of comparable pre-trained models for 3D shapes, recent methods utilize language-image pre-training to realize zero-shot 3D shape recognition. However, due to the modality gap, pretrained language-image models are not confident enough in the generalization to 3D shape recognition. Consequently, this paper aims to improve the confidence with view selection and hierarchical prompts. Leveraging the CLIP model as an example, we employ view selection on the vision side by identifying views with high prediction confidence from multiple rendered views of a 3D shape. On the textual side, the strategy of hierarchical prompts is proposed for the first time. The first layer prompts several classification candidates with traditional class-level descriptions, while the second layer refines the prediction based on function-level descriptions or further distinctions between the candidates. Remarkably, without the need for additional training, our proposed method achieves impressive zero-shot 3D classification accuracies of 84.44%, 91.51%, and 66.17% on ModelNet40, ModelNet10, and ShapeNet Core55, respectively. Furthermore, we will make the code publicly available to facilitate reproducibility and further research in this area.
Deep learning (DL) has become one of the mainstream and effective methods for point cloud analysis tasks such as detection, segmentation and classification. To reduce overfitting during training DL models and improve model performance especially when the amount and/or diversity of training data are limited, augmentation is often crucial. Although various point cloud data augmentation methods have been widely used in different point cloud processing tasks, there are currently no published systematic surveys or reviews of these methods. Therefore, this article surveys these methods, categorizing them into a taxonomy framework that comprises basic and specialized point cloud data augmentation methods. Through a comprehensive evaluation of these augmentation methods, this article identifies their potentials and limitations, serving as a useful reference for choosing appropriate augmentation methods. In addition, potential directions for future research are recommended. This survey contributes to providing a holistic overview of the current state of point cloud data augmentation, promoting its wider application and development.
The credibility of Artificial Intelligence (AI) models in medical imaging, particularly during the COVID-19 pandemic, has been challenged by reproducibility issues and obscured clinical insights. To address these concerns, we propose a Virtual Imaging Trials (VIT) framework, utilizing both clinical and simulated datasets to evaluate AI systems. This study focuses on using convolutional neural networks (CNNs) for COVID-19 diagnosis using computed tomography (CT) and chest radiography (CXR). We developed and tested multiple AI models, 3D ResNet-like and 2D EfficientNetv2 architectures, across diverse datasets. Our evaluation metrics included the area under the curve (AUC). Statistical analyses, such as the DeLong method for AUC confidence intervals, were employed to assess performance differences. Our findings demonstrate that VIT provides a robust platform for objective assessment, revealing significant influences of dataset characteristics, patient factors, and imaging physics on AI efficacy. Notably, models trained on the most diverse datasets showed the highest external testing performance, with AUC values ranging from 0.73 to 0.76 for CT and 0.70 to 0.73 for CXR. Internal testing yielded higher AUC values (0.77 to 0.85 for CT and 0.77 to 1.0 for CXR), highlighting a substantial drop in performance during external validation, which underscores the importance of diverse and comprehensive training and testing data. This approach enhances model transparency and reliability, offering nuanced insights into the factors driving AI performance and bridging the gap between experimental and clinical settings. The study underscores the potential of VIT to improve the reproducibility and clinical relevance of AI systems in medical imaging.
Though the success of CLIP-based training recipes in vision-language models, their scalability to more modalities (e.g., 3D, audio, etc.) is limited to large-scale data, which is expensive or even inapplicable for rare modalities. In this paper, we present ViT-Lens that facilitates efficient omni-modal representation learning by perceiving novel modalities with a pretrained ViT and aligning to a pre-defined space. Specifically, the modality-specific lens is tuned to project multimodal signals to the shared embedding space, which are then processed by a strong ViT that carries pre-trained image knowledge. The encoded multimodal representations are optimized toward aligning with the modal-independent space, pre-defined by off-the-shelf foundation models. A well-trained lens with a ViT backbone has the potential to serve as one of these foundation models, supervising the learning of subsequent modalities. ViT-Lens provides a unified solution for representation learning of increasing modalities with two appealing benefits: (i) Exploiting the pretrained ViT across tasks and domains effectively with efficient data regime; (ii) Emergent downstream capabilities of novel modalities are demonstrated due to the modality alignment space. We evaluate ViT-Lens in the context of 3D as an initial verification. In zero-shot 3D classification, ViT-Lens achieves substantial improvements over previous state-of-the-art, showing 52.0% accuracy on Objaverse-LVIS, 87.4% on ModelNet40, and 60.6% on ScanObjectNN. Furthermore, we enable zero-shot 3D question-answering by simply integrating the trained 3D lens into the InstructBLIP model without any adaptation. We will release the results of ViT-Lens on more modalities in the near future.
The ability to cope with out-of-distribution (OOD) corruptions and adversarial attacks is crucial in real-world safety-demanding applications. In this study, we develop a general mechanism to increase neural network robustness based on focus analysis. Recent studies have revealed the phenomenon of \textit{Overfocusing}, which leads to a performance drop. When the network is primarily influenced by small input regions, it becomes less robust and prone to misclassify under noise and corruptions. However, quantifying overfocusing is still vague and lacks clear definitions. Here, we provide a mathematical definition of \textbf{focus}, \textbf{overfocusing} and \textbf{underfocusing}. The notions are general, but in this study, we specifically investigate the case of 3D point clouds. We observe that corrupted sets result in a biased focus distribution compared to the clean training set. We show that as focus distribution deviates from the one learned in the training phase - classification performance deteriorates. We thus propose a parameter-free \textbf{refocusing} algorithm that aims to unify all corruptions under the same distribution. We validate our findings on a 3D zero-shot classification task, achieving SOTA in robust 3D classification on ModelNet-C dataset, and in adversarial defense against Shape-Invariant attack. Code is available in: https://github.com/yossilevii100/refocusing.
Due to the extremely low signal-to-noise ratio (SNR) and unknown poses (projection angles and image shifts) in cryo-electron microscopy (cryo-EM) experiments, reconstructing 3D volumes from 2D images is very challenging. In addition to these challenges, heterogeneous cryo-EM reconstruction requires conformational classification. In popular cryo-EM reconstruction algorithms, poses and conformation classification labels must be predicted for every input cryo-EM image, which can be computationally costly for large datasets. An emerging class of methods adopted the amortized inference approach. In these methods, only a subset of the input dataset is needed to train neural networks for the estimation of poses and conformations. Once trained, these neural networks can make pose/conformation predictions and 3D reconstructions at low cost for the entire dataset during inference. Unfortunately, when facing heterogeneous reconstruction tasks, it is hard for current amortized-inference-based methods to effectively estimate the conformational distribution and poses from entangled latent variables. Here, we propose a self-supervised variational autoencoder architecture called "HetACUMN" based on amortized inference. We employed an auxiliary conditional pose prediction task by inverting the order of encoder-decoder to explicitly enforce the disentanglement of conformation and pose predictions. Results on simulated datasets show that HetACUMN generated more accurate conformational classifications than other amortized or non-amortized methods. Furthermore, we show that HetACUMN is capable of performing heterogeneous 3D reconstructions of a real experimental dataset.
In recent years, 3D understanding has turned to 2D vision-language pre-trained models to overcome data scarcity challenges. However, existing methods simply transfer 2D alignment strategies, aligning 3D representations with single-view 2D images and coarse-grained parent category text. These approaches introduce information degradation and insufficient synergy issues, leading to performance loss. Information degradation arises from overlooking the fact that a 3D representation should be equivalent to a series of multi-view images and more fine-grained subcategory text. Insufficient synergy neglects the idea that a robust 3D representation should align with the joint vision-language space, rather than independently aligning with each modality. In this paper, we propose a multi-view joint modality modeling approach, termed JM3D, to obtain a unified representation for point cloud, text, and image. Specifically, a novel Structured Multimodal Organizer (SMO) is proposed to address the information degradation issue, which introduces contiguous multi-view images and hierarchical text to enrich the representation of vision and language modalities. A Joint Multi-modal Alignment (JMA) is designed to tackle the insufficient synergy problem, which models the joint modality by incorporating language knowledge into the visual modality. Extensive experiments on ModelNet40 and ScanObjectNN demonstrate the effectiveness of our proposed method, JM3D, which achieves state-of-the-art performance in zero-shot 3D classification. JM3D outperforms ULIP by approximately 4.3% on PointMLP and achieves an improvement of up to 6.5% accuracy on PointNet++ in top-1 accuracy for zero-shot 3D classification on ModelNet40. The source code and trained models for all our experiments are publicly available at https://github.com/Mr-Neko/JM3D.
Malnutrition poses a significant threat to global health, resulting from an inadequate intake of essential nutrients that adversely impacts vital organs and overall bodily functioning. Periodic examinations and mass screenings, incorporating both conventional and non-invasive techniques, have been employed to combat this challenge. However, these approaches suffer from critical limitations, such as the need for additional equipment, lack of comprehensive feature representation, absence of suitable health indicators, and the unavailability of smartphone implementations for precise estimations of Body Fat Percentage (BFP), Basal Metabolic Rate (BMR), and Body Mass Index (BMI) to enable efficient smart-malnutrition monitoring. To address these constraints, this study presents a groundbreaking, scalable, and robust smart malnutrition-monitoring system that leverages a single full-body image of an individual to estimate height, weight, and other crucial health parameters within a multi-modal learning framework. Our proposed methodology involves the reconstruction of a highly precise 3D point cloud, from which 512-dimensional feature embeddings are extracted using a headless-3D classification network. Concurrently, facial and body embeddings are also extracted, and through the application of learnable parameters, these features are then utilized to estimate weight accurately. Furthermore, essential health metrics, including BMR, BFP, and BMI, are computed to conduct a comprehensive analysis of the subject's health, subsequently facilitating the provision of personalized nutrition plans. While being robust to a wide range of lighting conditions across multiple devices, our model achieves a low Mean Absolute Error (MAE) of $\pm$ 4.7 cm and $\pm$ 5.3 kg in estimating height and weight.
Large pre-trained models have had a significant impact on computer vision by enabling multi-modal learning, where the CLIP model has achieved impressive results in image classification, object detection, and semantic segmentation. However, the model's performance on 3D point cloud processing tasks is limited due to the domain gap between depth maps from 3D projection and training images of CLIP. This paper proposes DiffCLIP, a new pre-training framework that incorporates stable diffusion with ControlNet to minimize the domain gap in the visual branch. Additionally, a style-prompt generation module is introduced for few-shot tasks in the textual branch. Extensive experiments on the ModelNet10, ModelNet40, and ScanObjectNN datasets show that DiffCLIP has strong abilities for 3D understanding. By using stable diffusion and style-prompt generation, DiffCLIP achieves an accuracy of 43.2\% for zero-shot classification on OBJ\_BG of ScanObjectNN, which is state-of-the-art performance, and an accuracy of 80.6\% for zero-shot classification on ModelNet10, which is comparable to state-of-the-art performance.
Recent advancements in multimodal pre-training have shown promising efficacy in 3D representation learning by aligning multimodal features across 3D shapes, their 2D counterparts, and language descriptions. However, the methods used by existing frameworks to curate such multimodal data, in particular language descriptions for 3D shapes, are not scalable, and the collected language descriptions are not diverse. To address this, we introduce ULIP-2, a simple yet effective tri-modal pre-training framework that leverages large multimodal models to automatically generate holistic language descriptions for 3D shapes. It only needs 3D data as input, eliminating the need for any manual 3D annotations, and is therefore scalable to large datasets. ULIP-2 is also equipped with scaled-up backbones for better multimodal representation learning. We conduct experiments on two large-scale 3D datasets, Objaverse and ShapeNet, and augment them with tri-modal datasets of 3D point clouds, images, and language for training ULIP-2. Experiments show that ULIP-2 demonstrates substantial benefits in three downstream tasks: zero-shot 3D classification, standard 3D classification with fine-tuning, and 3D captioning (3D-to-language generation). It achieves a new SOTA of 50.6% (top-1) on Objaverse-LVIS and 84.7% (top-1) on ModelNet40 in zero-shot classification. In the ScanObjectNN benchmark for standard fine-tuning, ULIP-2 reaches an overall accuracy of 91.5% with a compact model of only 1.4 million parameters. ULIP-2 sheds light on a new paradigm for scalable multimodal 3D representation learning without human annotations and shows significant improvements over existing baselines. The code and datasets are released at https://github.com/salesforce/ULIP.
We present view-synthesis autoencoders (VSA) in this paper, which is a self-supervised learning framework designed for vision transformers. Different from traditional 2D pretraining methods, VSA can be pre-trained with multi-view data. In each iteration, the input to VSA is one view (or multiple views) of a 3D object and the output is a synthesized image in another target pose. The decoder of VSA has several cross-attention blocks, which use the source view as value, source pose as key, and target pose as query. They achieve cross-attention to synthesize the target view. This simple approach realizes large-angle view synthesis and learns spatial invariant representation, where the latter is decent initialization for transformers on downstream tasks, such as 3D classification on ModelNet40, ShapeNet Core55, and ScanObjectNN. VSA outperforms existing methods significantly for linear probing and is competitive for fine-tuning. The code will be made publicly available.
Real-scanned point clouds are often incomplete due to viewpoint, occlusion, and noise, which hampers 3D geometric modeling and perception. Existing point cloud completion methods tend to generate global shape skeletons and hence lack fine local details. Furthermore, they mostly learn a deterministic partial-to-complete mapping, but overlook structural relations in man-made objects. To tackle these challenges, this paper proposes a variational framework, Variational Relational point Completion Network (VRCNet) with two appealing properties: 1) Probabilistic Modeling. In particular, we propose a dual-path architecture to enable principled probabilistic modeling across partial and complete clouds. One path consumes complete point clouds for reconstruction by learning a point VAE. The other path generates complete shapes for partial point clouds, whose embedded distribution is guided by distribution obtained from the reconstruction path during training. 2) Relational Enhancement. Specifically, we carefully design point self-attention kernel and point selective kernel module to exploit relational point features, which refines local shape details conditioned on the coarse completion. In addition, we contribute multi-view partial point cloud datasets (MVP and MVP-40 dataset) containing over 200,000 high-quality scans, which render partial 3D shapes from 26 uniformly distributed camera poses for each 3D CAD model. Extensive experiments demonstrate that VRCNet outperforms state-of-the-art methods on all standard point cloud completion benchmarks. Notably, VRCNet shows great generalizability and robustness on real-world point cloud scans. Moreover, we can achieve robust 3D classification for partial point clouds with the help of VRCNet, which can highly increase classification accuracy.
The assessment of aortic valve pathology using magnetic resonance imaging (MRI) typically relies on blood velocity estimates acquired using phase contrast (PC) MRI. However, abnormalities in blood flow through the aortic valve often manifest by the dephasing of blood signal in gated balanced steady-state free precession (bSSFP) scans (Cine MRI). We propose a 3D classification neural network (NN) to automatically identify aortic valve pathology (aortic regurgitation, aortic stenosis, mixed valve disease) from Cine MR images. We train and test our approach on a retrospective clinical dataset from three UK hospitals, using single-slice 3-chamber cine MRI from N = 576 patients. Our classification model accurately predicts the presence of aortic valve pathology (AVD) with an accuracy of 0.85 +/- 0.03 and can also correctly discriminate the type of AVD pathology (accuracy: 0.75 +/- 0.03). Gradient-weighted class activation mapping (Grad-CAM) confirms that the blood pool voxels close to the aortic root contribute the most to the classification. Our approach can be used to improve the diagnosis of AVD and optimise clinical CMR protocols for accurate and efficient AVD detection.
Adaptive radiotherapy is a growing field of study in cancer treatment due to it's objective in sparing healthy tissue. The standard of care in several institutions includes longitudinal cone-beam computed tomography (CBCT) acquisitions to monitor changes, but have yet to be used to improve tumor control while managing side-effects. The aim of this study is to demonstrate the clinical value of pre-treatment CBCT acquired daily during radiation therapy treatment for head and neck cancers for the downstream task of predicting severe toxicity occurrence: reactive feeding tube (NG), hospitalization and radionecrosis. For this, we propose a deformable 3D classification pipeline that includes a component analyzing the Jacobian matrix of the deformation between planning CT and longitudinal CBCT, as well as clinical data. The model is based on a multi-branch 3D residual convolutional neural network, while the CT to CBCT registration is based on a pair of VoxelMorph architectures. Accuracies of 85.8% and 75.3% was found for radionecrosis and hospitalization, respectively, with similar performance as early as after the first week of treatment. For NG tube risk, performance improves with increasing the timing of the CBCT fraction, reaching 83.1% after the $5_{th}$ week of treatment.
Multi-view projection techniques have shown themselves to be highly effective in achieving top-performing results in the recognition of 3D shapes. These methods involve learning how to combine information from multiple view-points. However, the camera view-points from which these views are obtained are often fixed for all shapes. To overcome the static nature of current multi-view techniques, we propose learning these view-points. Specifically, we introduce the Multi-View Transformation Network (MVTN), which uses differentiable rendering to determine optimal view-points for 3D shape recognition. As a result, MVTN can be trained end-to-end with any multi-view network for 3D shape classification. We integrate MVTN into a novel adaptive multi-view pipeline that is capable of rendering both 3D meshes and point clouds. Our approach demonstrates state-of-the-art performance in 3D classification and shape retrieval on several benchmarks (ModelNet40, ScanObjectNN, ShapeNet Core55). Further analysis indicates that our approach exhibits improved robustness to occlusion compared to other methods. We also investigate additional aspects of MVTN, such as 2D pretraining and its use for segmentation. To support further research in this area, we have released MVTorch, a PyTorch library for 3D understanding and generation using multi-view projections.
The recognition capabilities of current state-of-the-art 3D models are limited by datasets with a small number of annotated data and a pre-defined set of categories. In its 2D counterpart, recent advances have shown that similar problems can be significantly alleviated by employing knowledge from other modalities, such as language. Inspired by this, leveraging multimodal information for 3D modality could be promising to improve 3D understanding under the restricted data regime, but this line of research is not well studied. Therefore, we introduce ULIP to learn a unified representation of images, texts, and 3D point clouds by pre-training with object triplets from the three modalities. To overcome the shortage of training triplets, ULIP leverages a pre-trained vision-language model that has already learned a common visual and textual space by training with massive image-text pairs. Then, ULIP learns a 3D representation space aligned with the common image-text space, using a small number of automatically synthesized triplets. ULIP is agnostic to 3D backbone networks and can easily be integrated into any 3D architecture. Experiments show that ULIP effectively improves the performance of multiple recent 3D backbones by simply pre-training them on ShapeNet55 using our framework, achieving state-of-the-art performance in both standard 3D classification and zero-shot 3D classification on ModelNet40 and ScanObjectNN. ULIP also improves the performance of PointMLP by around 3% in 3D classification on ScanObjectNN, and outperforms PointCLIP by 28.8% on top-1 accuracy for zero-shot 3D classification on ModelNet40. Our code and pre-trained models are released at https://github.com/salesforce/ULIP.
With advances in deep learning model training strategies, the training of Point cloud classification methods is significantly improving. For example, PointNeXt, which adopts prominent training techniques and InvResNet layers into PointNet++, achieves over 7% improvement on the real-world ScanObjectNN dataset. However, most of these models use point coordinates features of neighborhood points mapped to higher dimensional space while ignoring the neighborhood point features computed before feeding to the network layers. In this paper, we revisit the PointNeXt model to study the usage and benefit of such neighborhood point features. We train and evaluate PointNeXt on ModelNet40 (synthetic), ScanObjectNN (real-world), and a recent large-scale, real-world grocery dataset, i.e., 3DGrocery100. In addition, we provide an additional inference strategy of weight averaging the top two checkpoints of PointNeXt to improve classification accuracy. Together with the abovementioned ideas, we gain 0.5%, 1%, 4.8%, 3.4%, and 1.6% overall accuracy on the PointNeXt model with real-world datasets, ScanObjectNN (hardest variant), 3DGrocery100's Apple10, Fruits, Vegetables, and Packages subsets, respectively. We also achieve a comparable 0.2% accuracy gain on ModelNet40.
Existing multi-view-based point cloud classification methods only utilize multiple views of point clouds and discard the point clouds from further processing. Among these methods, the Simple View model demonstrates that features from six orthogonal perspective projections of a point cloud achieved comparable 3D classification. However, points on the local structures overlap in these projections resulting in the loss of structural information. Also, we found that the performance of Simple View degrades at lower projection resolutions. We propose the use of neighbor projections along with object projections to learn finer local structural information. In this paper, we introduce SimpleView++ to concatenate features from the combined orthogonal perspective projections at object and neighbor levels with encoded features from the point cloud. We evaluated Simple-View++ using ModelNet40 and ScanObjectNN benchmark datasets. Our idea to use neighbor views broadly applies to other existing view-based methods. For example, neighbor views improve the results of MVTN by 2% on the hardest variant of ScanObjectNN. Our experiments also validate that neighbor view features considerably enhance classification accuracy at lower resolutions.
Large-scale pre-trained models have shown promising open-world performance for both vision and language tasks. However, their transferred capacity on 3D point clouds is still limited and only constrained to the classification task. In this paper, we first collaborate CLIP and GPT to be a unified 3D open-world learner, named as PointCLIP V2, which fully unleashes their potential for zero-shot 3D classification, segmentation, and detection. To better align 3D data with the pre-trained language knowledge, PointCLIP V2 contains two key designs. For the visual end, we prompt CLIP via a shape projection module to generate more realistic depth maps, narrowing the domain gap between projected point clouds with natural images. For the textual end, we prompt the GPT model to generate 3D-specific text as the input of CLIP's textual encoder. Without any training in 3D domains, our approach significantly surpasses PointCLIP by +42.90%, +40.44%, and +28.75% accuracy on three datasets for zero-shot 3D classification. On top of that, V2 can be extended to few-shot 3D classification, zero-shot 3D part segmentation, and 3D object detection in a simple manner, demonstrating our generalization ability for unified 3D open-world learning.
Despite recent success of self-supervised based contrastive learning model for 3D point clouds representation, the adversarial robustness of such pre-trained models raised concerns. Adversarial contrastive learning (ACL) is considered an effective way to improve the robustness of pre-trained models. In contrastive learning, the projector is considered an effective component for removing unnecessary feature information during contrastive pretraining and most ACL works also use contrastive loss with projected feature representations to generate adversarial examples in pretraining, while "unprojected " feature representations are used in generating adversarial inputs during inference.Because of the distribution gap between projected and "unprojected" features, their models are constrained of obtaining robust feature representations for downstream tasks. We introduce a new method to generate high-quality 3D adversarial examples for adversarial training by utilizing virtual adversarial loss with "unprojected" feature representations in contrastive learning framework. We present our robust aware loss function to train self-supervised contrastive learning framework adversarially. Furthermore, we find selecting high difference points with the Difference of Normal (DoN) operator as additional input for adversarial self-supervised contrastive learning can significantly improve the adversarial robustness of the pre-trained model. We validate our method, PointACL on downstream tasks, including 3D classification and 3D segmentation with multiple datasets. It obtains comparable robust accuracy over state-of-the-art contrastive adversarial learning methods.