Search

This paper presents ThinkDiff, a novel alignment paradigm that empowers text-to-image diffusion models with multimodal in-context understanding and reasoning capabilities by integrating the strengths of vision-language models (VLMs). Existing multimodal diffusion finetuning methods largely focus on pixel-level reconstruction rather than in-context reasoning, and are constrained by the complexity and limited availability of reasoning-based datasets. ThinkDiff addresses these challenges by leveraging vision-language training as a proxy task, aligning VLMs with the decoder of an encoder-decoder large language model (LLM) instead of a diffusion decoder. This proxy task builds on the observation that the $\textbf{LLM decoder}$ shares the same input feature space with $\textbf{diffusion decoders}$ that use the corresponding $\textbf{LLM encoder}$ for prompt embedding. As a result, aligning VLMs with diffusion decoders can be simplified through alignment with the LLM decoder. Without complex training and datasets, ThinkDiff effectively unleashes understanding, reasoning, and composing capabilities in diffusion models. Experiments demonstrate that ThinkDiff significantly improves accuracy from 19.2% to 46.3% on the challenging CoBSAT benchmark for multimodal in-context reasoning generation, with only 5 hours of training on 4 A100 GPUs. Additionally, ThinkDiff demonstrates exceptional performance in composing multiple images and texts into logically coherent images. Project page: https://mizhenxing.github.io/ThinkDiff., Comment: Project page: https://mizhenxing.github.io/ThinkDiff, 19 pages, 14 figures

Human understanding and generation are critical for modeling digital humans and humanoid embodiments. Recently, Human-centric Foundation Models (HcFMs) inspired by the success of generalist models, such as large language and vision models, have emerged to unify diverse human-centric tasks into a single framework, surpassing traditional task-specific approaches. In this survey, we present a comprehensive overview of HcFMs by proposing a taxonomy that categorizes current approaches into four groups: (1) Human-centric Perception Foundation Models that capture fine-grained features for multi-modal 2D and 3D understanding. (2) Human-centric AIGC Foundation Models that generate high-fidelity, diverse human-related content. (3) Unified Perception and Generation Models that integrate these capabilities to enhance both human understanding and synthesis. (4) Human-centric Agentic Foundation Models that extend beyond perception and generation to learn human-like intelligence and interactive behaviors for humanoid embodied tasks. We review state-of-the-art techniques, discuss emerging challenges and future research directions. This survey aims to serve as a roadmap for researchers and practitioners working towards more robust, versatile, and intelligent digital human and embodiments modeling., Comment: 9 pages

We present QSteed, a quantum compilation system that can be deployed on real quantum computing devices and quantum computing clusters. It is designed to meet the challenges of effectively compiling quantum tasks and managing multiple quantum backends. The system integrates two core components: a quantum compiler and a quantum computing resource virtualization manager, both of which provide standardized interfaces. The resource manager models quantum chips into different abstract layers, including the real quantum processing unit (QPU), the standard QPU (StdQPU), the substructure QPU (SubQPU), and the virtual QPU (VQPU), and stores this information in a quantum computing resource virtualization database, thus realizing the unified management of quantum computing devices. The quantum compiler adopts a modular and extensible design, providing a flexible framework for customizing compilation optimization strategies. It provides hardware-aware compilation algorithms that account for quantum gate noise and qubit coupling structures. By selecting the most suitable computing resources from the VQPU library, the compiler maps quantum tasks to the optimal qubit regions of the target device. We validated the effectiveness of the QSteed on the superconducting devices of the Quafu quantum cloud computing cluster. The quantum computing resource virtualization management technology of QSteed and the flexible and extensible design of its compiler make it possible to achieve unified management and task compilation for backend devices of multiple physical systems such as neutral-atom and ion-trap., Comment: White paper for QSteed, with 7 figures and 18 pages

This paper tackles the problem of generalizable 3D-aware generation from monocular datasets, e.g., ImageNet. The key challenge of this task is learning a robust 3D-aware representation without multi-view or dynamic data, while ensuring consistent texture and geometry across different viewpoints. Although some baseline methods are capable of 3D-aware generation, the quality of the generated images still lags behind state-of-the-art 2D generation approaches, which excel in producing high-quality, detailed images. To address this severe limitation, we propose a novel feed-forward pipeline based on pixel-aligned Gaussian Splatting, coined as F3D-Gaus, which can produce more realistic and reliable 3D renderings from monocular inputs. In addition, we introduce a self-supervised cycle-consistent constraint to enforce cross-view consistency in the learned 3D representation. This training strategy naturally allows aggregation of multiple aligned Gaussian primitives and significantly alleviates the interpolation limitations inherent in single-view pixel-aligned Gaussian Splatting. Furthermore, we incorporate video model priors to perform geometry-aware refinement, enhancing the generation of fine details in wide-viewpoint scenarios and improving the model's capability to capture intricate 3D textures. Extensive experiments demonstrate that our approach not only achieves high-quality, multi-view consistent 3D-aware generation from monocular datasets, but also significantly improves training and inference efficiency., Comment: Project Page: https://w-ted.github.io/publications/F3D-Gaus

Diffusion-based human animation aims to animate a human character based on a source human image as well as driving signals such as a sequence of poses. Leveraging the generative capacity of diffusion model, existing approaches are able to generate high-fidelity poses, but struggle with significant viewpoint changes, especially in zoom-in/zoom-out scenarios where camera-character distance varies. This limits the applications such as cinematic shot type plan or camera control. We propose a pose-correlated reference selection diffusion network, supporting substantial viewpoint variations in human animation. Our key idea is to enable the network to utilize multiple reference images as input, since significant viewpoint changes often lead to missing appearance details on the human body. To eliminate the computational cost, we first introduce a novel pose correlation module to compute similarities between non-aligned target and source poses, and then propose an adaptive reference selection strategy, utilizing the attention map to identify key regions for animation generation. To train our model, we curated a large dataset from public TED talks featuring varied shots of the same character, helping the model learn synthesis for different perspectives. Our experimental results show that with the same number of reference images, our model performs favorably compared to the current SOTA methods under large viewpoint change. We further show that the adaptive reference selection is able to choose the most relevant reference regions to generate humans under free viewpoints., Comment: Under review; Project page: https://harlanhong.github.io/publications/fvhuman/index.html

Talking head video generation aims to generate a realistic talking head video that preserves the person's identity from a source image and the motion from a driving video. Despite the promising progress made in the field, it remains a challenging and critical problem to generate videos with accurate poses and fine-grained facial details simultaneously. Essentially, facial motion is often highly complex to model precisely, and the one-shot source face image cannot provide sufficient appearance guidance during generation due to dynamic pose changes. To tackle the problem, we propose to jointly learn motion and appearance codebooks and perform multi-scale codebook compensation to effectively refine both the facial motion conditions and appearance features for talking face image decoding. Specifically, the designed multi-scale motion and appearance codebooks are learned simultaneously in a unified framework to store representative global facial motion flow and appearance patterns. Then, we present a novel multi-scale motion and appearance compensation module, which utilizes a transformer-based codebook retrieval strategy to query complementary information from the two codebooks for joint motion and appearance compensation. The entire process produces motion flows of greater flexibility and appearance features with fewer distortions across different scales, resulting in a high-quality talking head video generation framework. Extensive experiments on various benchmarks validate the effectiveness of our approach and demonstrate superior generation results from both qualitative and quantitative perspectives when compared to state-of-the-art competitors., Comment: Project page: https://shaelynz.github.io/synergize-motion-appearance/

In recent years, simultaneous learning of multiple dense prediction tasks with partially annotated label data has emerged as an important research area. Previous works primarily focus on constructing cross-task consistency or conducting adversarial training to regularize cross-task predictions, which achieve promising performance improvements, while still suffering from the lack of direct pixel-wise supervision for multi-task dense predictions. To tackle this challenge, we propose a novel approach to optimize a set of learnable hierarchical task tokens, including global and fine-grained ones, to discover consistent pixel-wise supervision signals in both feature and prediction levels. Specifically, the global task tokens are designed for effective cross-task feature interactions in a global context. Then, a group of fine-grained task-specific spatial tokens for each task is learned from the corresponding global task tokens. It is embedded to have dense interactions with each task-specific feature map. The learned global and local fine-grained task tokens are further used to discover pseudo task-specific dense labels at different levels of granularity, and they can be utilized to directly supervise the learning of the multi-task dense prediction framework. Extensive experimental results on challenging NYUD-v2, Cityscapes, and PASCAL Context datasets demonstrate significant improvements over existing state-of-the-art methods for partially annotated multi-task dense prediction.

In NeRF, a critical problem is to effectively estimate the occupancy to guide empty-space skipping and point sampling. Grid-based methods work well for small-scale scenes. However, on large-scale scenes, they are limited by predefined bounding boxes, grid resolutions, and high memory usage for grid updates, and thus struggle to speed up training for large-scale, irregularly bounded and complex urban scenes without sacrificing accuracy. In this paper, we propose to learn a continuous and compact large-scale occupancy network, which can classify 3D points as occupied or unoccupied points. We train this occupancy network end-to-end together with the radiance field in a self-supervised manner by three designs. First, we propose a novel imbalanced occupancy loss to regularize the occupancy network. It makes the occupancy network effectively control the ratio of unoccupied and occupied points, motivated by the prior that most of 3D scene points are unoccupied. Second, we design an imbalanced architecture containing a large scene network and a small empty space network to separately encode occupied and unoccupied points classified by the occupancy network. This imbalanced structure can effectively model the imbalanced nature of occupied and unoccupied regions. Third, we design an explicit density loss to guide the occupancy network, making the density of unoccupied points smaller. As far as we know, we are the first to learn a continuous and compact occupancy of large-scale NeRF by a network. In our experiments, our occupancy network can quickly learn more compact, accurate and smooth occupancy compared to the occupancy grid. With our learned occupancy as guidance for empty space skipping on challenging large-scale benchmarks, our method consistently obtains higher accuracy compared to the occupancy grid, and our method can speed up state-of-the-art NeRF methods without sacrificing accuracy., Comment: 13 pages

Generating lifelike human motions from descriptive texts has experienced remarkable research focus in the recent years, propelled by the emerging requirements of digital humans.Despite impressive advances, existing approaches are often constrained by limited control modalities, task specificity, and focus solely on body motion representations.In this paper, we present MotionGPT-2, a unified Large Motion-Language Model (LMLM) that addresses these limitations. MotionGPT-2 accommodates multiple motion-relevant tasks and supporting multimodal control conditions through pre-trained Large Language Models (LLMs). It quantizes multimodal inputs-such as text and single-frame poses-into discrete, LLM-interpretable tokens, seamlessly integrating them into the LLM's vocabulary. These tokens are then organized into unified prompts, guiding the LLM to generate motion outputs through a pretraining-then-finetuning paradigm. We also show that the proposed MotionGPT-2 is highly adaptable to the challenging 3D holistic motion generation task, enabled by the innovative motion discretization framework, Part-Aware VQVAE, which ensures fine-grained representations of body and hand movements. Extensive experiments and visualizations validate the effectiveness of our method, demonstrating the adaptability of MotionGPT-2 across motion generation, motion captioning, and generalized motion completion tasks.

This research aims to comprehensively explore building a multimodal foundation model for egocentric video understanding. To achieve this goal, we work on three fronts. First, as there is a lack of QA data for egocentric video understanding, we develop a data engine that efficiently generates 7M high-quality QA samples for egocentric videos ranging from 30 seconds to one hour long, based on human-annotated data. This is currently the largest egocentric QA dataset. Second, we contribute a challenging egocentric QA benchmark with 629 videos and 7,026 questions to evaluate the models' ability in recognizing and memorizing visual details across videos of varying lengths. We introduce a new de-biasing evaluation method to help mitigate the unavoidable language bias present in the models being evaluated. Third, we propose a specialized multimodal architecture featuring a novel "Memory Pointer Prompting" mechanism. This design includes a global glimpse step to gain an overarching understanding of the entire video and identify key visual information, followed by a fallback step that utilizes the key visual information to generate responses. This enables the model to more effectively comprehend extended video content. With the data, benchmark, and model, we successfully build MM-Ego, an egocentric multimodal LLM that shows powerful performance on egocentric video understanding., Comment: Technical Report

Neural Radiance Fields (NeRF) are widely used for novel-view synthesis and have been adapted for 3D Object Detection (3DOD), offering a promising approach to 3DOD through view-synthesis representation. However, NeRF faces inherent limitations: (i) limited representational capacity for 3DOD due to its implicit nature, and (ii) slow rendering speeds. Recently, 3D Gaussian Splatting (3DGS) has emerged as an explicit 3D representation that addresses these limitations. Inspired by these advantages, this paper introduces 3DGS into 3DOD for the first time, identifying two main challenges: (i) Ambiguous spatial distribution of Gaussian blobs: 3DGS primarily relies on 2D pixel-level supervision, resulting in unclear 3D spatial distribution of Gaussian blobs and poor differentiation between objects and background, which hinders 3DOD; (ii) Excessive background blobs: 2D images often include numerous background pixels, leading to densely reconstructed 3DGS with many noisy Gaussian blobs representing the background, negatively affecting detection. To tackle the challenge (i), we leverage the fact that 3DGS reconstruction is derived from 2D images, and propose an elegant and efficient solution by incorporating 2D Boundary Guidance to significantly enhance the spatial distribution of Gaussian blobs, resulting in clearer differentiation between objects and their background. To address the challenge (ii), we propose a Box-Focused Sampling strategy using 2D boxes to generate object probability distribution in 3D spaces, allowing effective probabilistic sampling in 3D to retain more object blobs and reduce noisy background blobs. Benefiting from our designs, our 3DGS-DET significantly outperforms the SOTA NeRF-based method, NeRF-Det, achieving improvements of +6.6 on mAP@0.25 and +8.1 on mAP@0.5 for the ScanNet dataset, and impressive +31.5 on mAP@0.25 for the ARKITScenes dataset., Comment: Code Page: https://github.com/yangcaoai/3DGS-DET

Audio-driven talking head synthesis strives to generate lifelike video portraits from provided audio. The diffusion model, recognized for its superior quality and robust generalization, has been explored for this task. However, establishing a robust correspondence between temporal audio cues and corresponding spatial facial expressions with diffusion models remains a significant challenge in talking head generation. To bridge this gap, we present DreamHead, a hierarchical diffusion framework that learns spatial-temporal correspondences in talking head synthesis without compromising the model's intrinsic quality and adaptability.~DreamHead learns to predict dense facial landmarks from audios as intermediate signals to model the spatial and temporal correspondences.~Specifically, a first hierarchy of audio-to-landmark diffusion is first designed to predict temporally smooth and accurate landmark sequences given audio sequence signals. Then, a second hierarchy of landmark-to-image diffusion is further proposed to produce spatially consistent facial portrait videos, by modeling spatial correspondences between the dense facial landmark and appearance. Extensive experiments show that proposed DreamHead can effectively learn spatial-temporal consistency with the designed hierarchical diffusion and produce high-fidelity audio-driven talking head videos for multiple identities.

A quantum state for being an eigenstate of some local Hamiltonian should be constraint by zero energy variance and consequently, the constraint is rather strong that a single eigenstate may uniquely determine the Hamiltonian. For non-Hermitian systems, it is natural to expect that determining the Hamiltonian requires a pair of both left and right eigenstates. Here, we observe that it can be sufficient to determine a non-Hermitian Hamiltonian from a single right or left eigenstate. Our approach is based on the quantum covariance matrix, where the solution of Hamiltonian corresponds to the complex null vector. Our scheme favours non-Hermitian Hamiltonian learning on experimental quantum systems, as only the right eigenstates there can be accessed. Furthermore, we use numerical simulations to examine the effects of measurement errors and show the stability of our scheme.

In ground-based astronomy, the ability to couple light into single-mode fibers (SMFs) is limited by atmospheric turbulence, which prohibits the use of many astrophotonic instruments. We propose a silicon-on-insulator photonic chip capable of coherently coupling the out-of-phase beamlets from the subapertures of a telescope pupil into an SMF. The photonic integrated circuit (PIC) consists of an array of grating couplers that are used to inject light from free space into single-mode waveguides on the chip. Metallic heaters modulate the refractive index of a coiled section of the waveguides, facilitating the co-phasing of the propagating modes. The phased beamlets can then be coherently combined to efficiently deliver the light to an output SMF. In an adaptive optics (AO) system, the phase corrector acts as a deformable mirror (DM) commanded by a controller that takes phase measurements from a wavefront sensor (WFS). We present experimental results for the PIC tested on an AO testbed and compare the performance to simulations., Comment: 13 pages, 11 figures, 2 tables, submitted to SPIE Adaptive Optics Systems IX

We present an analytical model of integrable turbulence in the focusing nonlinear Schr\"odinger (fNLS) equation, generated by a one-parameter family of finite-band elliptic potentials in the semiclassical limit. We show that the spectrum of these potentials exhibits a thermodynamic band/gap scaling compatible with that of soliton and breather gases depending on the value of the elliptic parameter m of the potential. We then demonstrate that, upon augmenting the potential by a small random noise (which is inevitably present in real physical systems), the solution of the fNLS equation evolves into a fully randomized, spatially homogeneous breather gas, a phenomenon we call breather gas fission. We show that the statistical properties of the breather gas at large times are determined by the spectral density of states generated by the unperturbed initial potential. We analytically compute the kurtosis of the breather gas as a function of the elliptic parameter m, and we show that it is greater than 2 for all non-zero m, implying non-Gaussian statistics. Finally, we verify the theoretical predictions by comparison with direct numerical simulations of the fNLS equation. These results establish a link between semiclassical limits of integrable systems and the statistical characterization of their soliton and breather gases., Comment: 11 pages, 11 figures

This paper introduces Motion-oriented Compositional Neural Radiance Fields (MoCo-NeRF), a framework designed to perform free-viewpoint rendering of monocular human videos via novel non-rigid motion modeling approach. In the context of dynamic clothed humans, complex cloth dynamics generate non-rigid motions that are intrinsically distinct from skeletal articulations and critically important for the rendering quality. The conventional approach models non-rigid motions as spatial (3D) deviations in addition to skeletal transformations. However, it is either time-consuming or challenging to achieve optimal quality due to its high learning complexity without a direct supervision. To target this problem, we propose a novel approach of modeling non-rigid motions as radiance residual fields to benefit from more direct color supervision in the rendering and utilize the rigid radiance fields as a prior to reduce the complexity of the learning process. Our approach utilizes a single multiresolution hash encoding (MHE) to concurrently learn the canonical T-pose representation from rigid skeletal motions and the radiance residual field for non-rigid motions. Additionally, to further improve both training efficiency and usability, we extend MoCo-NeRF to support simultaneous training of multiple subjects within a single framework, thanks to our effective design for modeling non-rigid motions. This scalability is achieved through the integration of a global MHE and learnable identity codes in addition to multiple local MHEs. We present extensive results on ZJU-MoCap and MonoCap, clearly demonstrating state-of-the-art performance in both single- and multi-subject settings. The code and model will be made publicly available at the project page: https://stevejaehyeok.github.io/publications/moco-nerf., Comment: Accepted by ECCV2024

Optical beams and starlight distorted by atmospheric turbulence can be corrected with adaptive optics systems to enable efficient coupling into single-mode fibers. Deformable mirrors, used to flatten the wavefront in astronomical telescopes, are costly, sensitive, and complex mechanical components that require careful calibration to enable high-quality imaging in astronomy, microscopy, and vision science. They are also impractical to deploy in large numbers for non-imaging applications like free-space optical communication. Here, we propose a photonic integrated c rcuit capable of spatially sampling the wavefront collected by the telescope and co-phasing the subapertures to maximize the flux delivered to an output single-mode fiber as the integrated photonic implementation of a deformable mirror. We present the results of end-to-end simulations to quantify the performance of the proposed photonic solution under varying atmospheric conditions toward realizing an adaptive optics system without a deformable mirror for free-space optical receivers., Comment: 14 pages, 8 figures, 1 table, to be published in Optics Express. Corresponding author: Momen Diab (momen.diab@utoronto.ca)

One-shot talking head video generation uses a source image and driving video to create a synthetic video where the source person's facial movements imitate those of the driving video. However, differences in scale between the source and driving images remain a challenge for face reenactment. Existing methods attempt to locate a frame in the driving video that aligns best with the source image, but imprecise alignment can result in suboptimal outcomes. To this end, we introduce a scale transformation module that can automatically adjust the scale of the driving image to fit that of the source image, by using the information of scale difference maintained in the detected keypoints of the source image and the driving frame. Furthermore, to keep perceiving the scale information of faces during the generation process, we incorporate the scale information learned from the scale transformation module into each layer of the generation process to produce a final result with an accurate scale. Our method can perform accurate motion transfer between the two images without any anchor frame, achieved through the contributions of the proposed online scale transformation facial reenactment network. Extensive experiments have demonstrated that our proposed method adjusts the scale of the driving face automatically according to the source face, and generates high-quality faces with an accurate scale in the cross-identity facial reenactment.

Recent advancements in autonomous driving technologies involve the capability to effectively process and learn from extensive real-world driving data. Current imitation learning and offline reinforcement learning methods have shown remarkable promise in autonomous systems, harnessing the power of offline datasets to make informed decisions in open-loop (non-reactive agents) settings. However, learning-based agents face significant challenges when transferring knowledge from open-loop to closed-loop (reactive agents) environment. The performance is significantly impacted by data distribution shift, sample efficiency, the complexity of uncovering hidden world models and physics. To address these issues, we propose Sample-efficient Imitative Multi-token Decision Transformer (SimDT). SimDT introduces multi-token prediction, online imitative learning pipeline and prioritized experience replay to sequence-modelling reinforcement learning. The performance is evaluated through empirical experiments and results exceed popular imitation and reinforcement learning algorithms both in open-loop and closed-loop settings on Waymax benchmark. SimDT exhibits 41% reduction in collision rate and 18% improvement in reaching the destination compared with the baseline method.

In this paper, we introduce a novel path to $\textit{general}$ human motion generation by focusing on 2D space. Traditional methods have primarily generated human motions in 3D, which, while detailed and realistic, are often limited by the scope of available 3D motion data in terms of both the size and the diversity. To address these limitations, we exploit extensive availability of 2D motion data. We present $\textbf{Holistic-Motion2D}$, the first comprehensive and large-scale benchmark for 2D whole-body motion generation, which includes over 1M in-the-wild motion sequences, each paired with high-quality whole-body/partial pose annotations and textual descriptions. Notably, Holistic-Motion2D is ten times larger than the previously largest 3D motion dataset. We also introduce a baseline method, featuring innovative $\textit{whole-body part-aware attention}$ and $\textit{confidence-aware modeling}$ techniques, tailored for 2D $\underline{\text T}$ext-driv$\underline{\text{EN}}$ whole-bo$\underline{\text D}$y motion gen$\underline{\text{ER}}$ation, namely $\textbf{Tender}$. Extensive experiments demonstrate the effectiveness of $\textbf{Holistic-Motion2D}$ and $\textbf{Tender}$ in generating expressive, diverse, and realistic human motions. We also highlight the utility of 2D motion for various downstream applications and its potential for lifting to 3D motion. The page link is: https://holistic-motion2d.github.io., Comment: 22 pages, 11figures, 17 tables

The advancement of diffusion models has pushed the boundary of text-to-3D object generation. While it is straightforward to composite objects into a scene with reasonable geometry, it is nontrivial to texture such a scene perfectly due to style inconsistency and occlusions between objects. To tackle these problems, we propose a coarse-to-fine 3D scene texturing framework, referred to as RoomTex, to generate high-fidelity and style-consistent textures for untextured compositional scene meshes. In the coarse stage, RoomTex first unwraps the scene mesh to a panoramic depth map and leverages ControlNet to generate a room panorama, which is regarded as the coarse reference to ensure the global texture consistency. In the fine stage, based on the panoramic image and perspective depth maps, RoomTex will refine and texture every single object in the room iteratively along a series of selected camera views, until this object is completely painted. Moreover, we propose to maintain superior alignment between RGB and depth spaces via subtle edge detection methods. Extensive experiments show our method is capable of generating high-quality and diverse room textures, and more importantly, supporting interactive fine-grained texture control and flexible scene editing thanks to our inpainting-based framework and compositional mesh input. Our project page is available at https://qwang666.github.io/RoomTex/.

Open-vocabulary 3D Object Detection (OV-3DDet) addresses the detection of objects from an arbitrary list of novel categories in 3D scenes, which remains a very challenging problem. In this work, we propose CoDAv2, a unified framework designed to innovatively tackle both the localization and classification of novel 3D objects, under the condition of limited base categories. For localization, the proposed 3D Novel Object Discovery (3D-NOD) strategy utilizes 3D geometries and 2D open-vocabulary semantic priors to discover pseudo labels for novel objects during training. 3D-NOD is further extended with an Enrichment strategy that significantly enriches the novel object distribution in the training scenes, and then enhances the model's ability to localize more novel objects. The 3D-NOD with Enrichment is termed 3D-NODE. For classification, the Discovery-driven Cross-modal Alignment (DCMA) module aligns features from 3D point clouds and 2D/textual modalities, employing both class-agnostic and class-specific alignments that are iteratively refined to handle the expanding vocabulary of objects. Besides, 2D box guidance boosts the classification accuracy against complex background noises, which is coined as Box-DCMA. Extensive evaluation demonstrates the superiority of CoDAv2. CoDAv2 outperforms the best-performing method by a large margin (AP_Novel of 9.17 vs. 3.61 on SUN-RGBD and 9.12 vs. 3.74 on ScanNetv2). Source code and pre-trained models are available at the GitHub project page., Comment: Code Page: https://github.com/yangcaoai/CoDA_NeurIPS2023 This paper has been submitted to IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) for possible publication