Search

Multimodal large language models (MLLMs) have demonstrated strong performance across various tasks without requiring training from scratch. However, they face significant computational and memory constraints, particularly when processing multimodal inputs that exceed context length, limiting their scalability. In this paper, we introduce a new approach, \textbf{TRSM} (\textbf{T}oken \textbf{R}eduction via \textbf{S}emantic \textbf{M}atch), which effectively reduces the number of visual tokens without compromising MLLM performance. Inspired by how humans process multimodal tasks, TRSM leverages semantic information from one modality to match relevant semantics in another, reducing the number of visual tokens.Specifically, to retain task relevant visual tokens, we use the text prompt as a query vector to retrieve the most similar vectors from the visual prompt and merge them with the text tokens. Based on experimental results, when applied to LLaVA-1.5\cite{liu2023}, our approach compresses the visual tokens by 20\%, achieving comparable performance across diverse visual question-answering and reasoning tasks., Comment: 8 pages, 2 figures,3 tables

In this paper, we present a method, VectorPD, for converting a given human face image into a vector portrait sketch. VectorPD supports different levels of abstraction by simply controlling the number of strokes. Since vector graphics are composed of different shape primitives, it is challenging for rendering complex faces to accurately express facial details and structure. To address this, VectorPD employs a novel two-round optimization mechanism. We first initialize the strokes with facial keypoints, and generate a basic portrait sketch by a CLIP-based Semantic Loss. Then we complete the face structure through VGG-based Structure Loss, and propose a novel Crop-based Shadow Loss to enrich the shadow details of the sketch, achieving a visually pleasing portrait sketch. Quantitative and qualitative evaluations both demonstrate that the portrait sketches generated by VectorPD can produce better visual effects than existing state-of-the-art methods, maintaining as much fidelity as possible at different levels of abstraction., Comment: 9 pages, 12 figures

Scene sketching is to convert a scene into a simplified, abstract representation that captures the essential elements and composition of the original scene. It requires semantic understanding of the scene and consideration of different regions within the scene. Since scenes often contain diverse visual information across various regions, such as foreground objects, background elements, and spatial divisions, dealing with these different regions poses unique difficulties. In this paper, we define a sketch as some sets of Bezier curves. We optimize the different regions of input scene in multiple rounds. In each round of optimization, strokes sampled from the next region can seamlessly be integrated into the sketch generated in the previous round of optimization. We propose additional stroke initialization method to ensure the integrity of the scene and the convergence of optimization. A novel CLIP-Based Semantic loss and a VGG-Based Feature loss are utilized to guide our multi-round optimization. Extensive experimental results on the quality and quantity of the generated sketches confirm the effectiveness of our method., Comment: 9 pages, 9 figures

In this paper, we present a new text-guided 3D shape generation approach DreamStone that uses images as a stepping stone to bridge the gap between text and shape modalities for generating 3D shapes without requiring paired text and 3D data. The core of our approach is a two-stage feature-space alignment strategy that leverages a pre-trained single-view reconstruction (SVR) model to map CLIP features to shapes: to begin with, map the CLIP image feature to the detail-rich 3D shape space of the SVR model, then map the CLIP text feature to the 3D shape space through encouraging the CLIP-consistency between rendered images and the input text. Besides, to extend beyond the generative capability of the SVR model, we design a text-guided 3D shape stylization module that can enhance the output shapes with novel structures and textures. Further, we exploit pre-trained text-to-image diffusion models to enhance the generative diversity, fidelity, and stylization capability. Our approach is generic, flexible, and scalable, and it can be easily integrated with various SVR models to expand the generative space and improve the generative fidelity. Extensive experimental results demonstrate that our approach outperforms the state-of-the-art methods in terms of generative quality and consistency with the input text. Codes and models are released at https://github.com/liuzhengzhe/DreamStone-ISS., Comment: IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)

This paper presents a new approach for 3D shape generation, inversion, and manipulation, through a direct generative modeling on a continuous implicit representation in wavelet domain. Specifically, we propose a compact wavelet representation with a pair of coarse and detail coefficient volumes to implicitly represent 3D shapes via truncated signed distance functions and multi-scale biorthogonal wavelets. Then, we design a pair of neural networks: a diffusion-based generator to produce diverse shapes in the form of the coarse coefficient volumes and a detail predictor to produce compatible detail coefficient volumes for introducing fine structures and details. Further, we may jointly train an encoder network to learn a latent space for inverting shapes, allowing us to enable a rich variety of whole-shape and region-aware shape manipulations. Both quantitative and qualitative experimental results manifest the compelling shape generation, inversion, and manipulation capabilities of our approach over the state-of-the-art methods., Comment: arXiv admin note: substantial text overlap with arXiv:2209.08725

This paper presents a new approach for 3D shape generation, enabling direct generative modeling on a continuous implicit representation in wavelet domain. Specifically, we propose a compact wavelet representation with a pair of coarse and detail coefficient volumes to implicitly represent 3D shapes via truncated signed distance functions and multi-scale biorthogonal wavelets, and formulate a pair of neural networks: a generator based on the diffusion model to produce diverse shapes in the form of coarse coefficient volumes; and a detail predictor to further produce compatible detail coefficient volumes for enriching the generated shapes with fine structures and details. Both quantitative and qualitative experimental results manifest the superiority of our approach in generating diverse and high-quality shapes with complex topology and structures, clean surfaces, and fine details, exceeding the 3D generation capabilities of the state-of-the-art models.

Text-guided 3D shape generation remains challenging due to the absence of large paired text-shape data, the substantial semantic gap between these two modalities, and the structural complexity of 3D shapes. This paper presents a new framework called Image as Stepping Stone (ISS) for the task by introducing 2D image as a stepping stone to connect the two modalities and to eliminate the need for paired text-shape data. Our key contribution is a two-stage feature-space-alignment approach that maps CLIP features to shapes by harnessing a pre-trained single-view reconstruction (SVR) model with multi-view supervisions: first map the CLIP image feature to the detail-rich shape space in the SVR model, then map the CLIP text feature to the shape space and optimize the mapping by encouraging CLIP consistency between the input text and the rendered images. Further, we formulate a text-guided shape stylization module to dress up the output shapes with novel textures. Beyond existing works on 3D shape generation from text, our new approach is general for creating shapes in a broad range of categories, without requiring paired text-shape data. Experimental results manifest that our approach outperforms the state-of-the-arts and our baselines in terms of fidelity and consistency with text. Further, our approach can stylize the generated shapes with both realistic and fantasy structures and textures., Comment: ICLR 2023 spotlight

This paper introduces a novel framework called DTNet for 3D mesh reconstruction and generation via Disentangled Topology. Beyond previous works, we learn a topology-aware neural template specific to each input then deform the template to reconstruct a detailed mesh while preserving the learned topology. One key insight is to decouple the complex mesh reconstruction into two sub-tasks: topology formulation and shape deformation. Thanks to the decoupling, DT-Net implicitly learns a disentangled representation for the topology and shape in the latent space. Hence, it can enable novel disentangled controls for supporting various shape generation applications, e.g., remix the topologies of 3D objects, that are not achievable by previous reconstruction works. Extensive experimental results demonstrate that our method is able to produce high-quality meshes, particularly with diverse topologies, as compared with the state-of-the-art methods.

This work presents an innovative method for point set self-embedding, that encodes the structural information of a dense point set into its sparser version in a visual but imperceptible form. The self-embedded point set can function as the ordinary downsampled one and be visualized efficiently on mobile devices. Particularly, we can leverage the self-embedded information to fully restore the original point set for detailed analysis on remote servers. This task is challenging since both the self-embedded point set and the restored point set should resemble the original one. To achieve a learnable self-embedding scheme, we design a novel framework with two jointly-trained networks: one to encode the input point set into its self-embedded sparse point set and the other to leverage the embedded information for inverting the original point set back. Further, we develop a pair of up-shuffle and down-shuffle units in the two networks, and formulate loss terms to encourage the shape similarity and point distribution in the results. Extensive qualitative and quantitative results demonstrate the effectiveness of our method on both synthetic and real-scanned datasets., Comment: Accepted by IEEE Transactions on Visualization and Computer Graphics (IEEE TVCG), 2022. All resources can be found at https://liruihui.github.io/

Point cloud upsampling is to densify a sparse point set acquired from 3D sensors, providing a denser representation for the underlying surface. Existing methods divide the input points into small patches and upsample each patch separately, however, ignoring the global spatial consistency between patches. In this paper, we present a novel method PC$^2$-PU, which explores patch-to-patch and point-to-point correlations for more effective and robust point cloud upsampling. Specifically, our network has two appealing designs: (i) We take adjacent patches as supplementary inputs to compensate the loss structure information within a single patch and introduce a Patch Correlation Module to capture the difference and similarity between patches. (ii) After augmenting each patch's geometry, we further introduce a Point Correlation Module to reveal the relationship of points inside each patch to maintain the local spatial consistency. Extensive experiments on both synthetic and real scanned datasets demonstrate that our method surpasses previous upsampling methods, particularly with the noisy inputs. The code and data are at \url{https://github.com/chenlongwhu/PC2-PU.git}., Comment: Accepted to ACM MM 2022

We present SP-GAN, a new unsupervised sphere-guided generative model for direct synthesis of 3D shapes in the form of point clouds. Compared with existing models, SP-GAN is able to synthesize diverse and high-quality shapes with fine details and promote controllability for part-aware shape generation and manipulation, yet trainable without any parts annotations. In SP-GAN, we incorporate a global prior (uniform points on a sphere) to spatially guide the generative process and attach a local prior (a random latent code) to each sphere point to provide local details. The key insight in our design is to disentangle the complex 3D shape generation task into a global shape modeling and a local structure adjustment, to ease the learning process and enhance the shape generation quality. Also, our model forms an implicit dense correspondence between the sphere points and points in every generated shape, enabling various forms of structure-aware shape manipulations such as part editing, part-wise shape interpolation, and multi-shape part composition, etc., beyond the existing generative models. Experimental results, which include both visual and quantitative evaluations, demonstrate that our model is able to synthesize diverse point clouds with fine details and less noise, as compared with the state-of-the-art models., Comment: SIGGRAPH 2021, website https://liruihui.github.io/publication/SP-GAN/

Point clouds produced by 3D scanning are often sparse, non-uniform, and noisy. Recent upsampling approaches aim to generate a dense point set, while achieving both distribution uniformity and proximity-to-surface, and possibly amending small holes, all in a single network. After revisiting the task, we propose to disentangle the task based on its multi-objective nature and formulate two cascaded sub-networks, a dense generator and a spatial refiner. The dense generator infers a coarse but dense output that roughly describes the underlying surface, while the spatial refiner further fine-tunes the coarse output by adjusting the location of each point. Specifically, we design a pair of local and global refinement units in the spatial refiner to evolve a coarse feature map. Also, in the spatial refiner, we regress a per-point offset vector to further adjust the coarse outputs in fine-scale. Extensive qualitative and quantitative results on both synthetic and real-scanned datasets demonstrate the superiority of our method over the state-of-the-arts., Comment: CVPR 2021, website https://liruihui.github.io/

This paper presents a deep normal filtering network, called DNF-Net, for mesh denoising. To better capture local geometry, our network processes the mesh in terms of local patches extracted from the mesh. Overall, DNF-Net is an end-to-end network that takes patches of facet normals as inputs and directly outputs the corresponding denoised facet normals of the patches. In this way, we can reconstruct the geometry from the denoised normals with feature preservation. Besides the overall network architecture, our contributions include a novel multi-scale feature embedding unit, a residual learning strategy to remove noise, and a deeply-supervised joint loss function. Compared with the recent data-driven works on mesh denoising, DNF-Net does not require manual input to extract features and better utilizes the training data to enhance its denoising performance. Finally, we present comprehensive experiments to evaluate our method and demonstrate its superiority over the state of the art on both synthetic and real-scanned meshes.

Recently, many deep neural networks were designed to process 3D point clouds, but a common drawback is that rotation invariance is not ensured, leading to poor generalization to arbitrary orientations. In this paper, we introduce a new low-level purely rotation-invariant representation to replace common 3D Cartesian coordinates as the network inputs. Also, we present a network architecture to embed these representations into features, encoding local relations between points and their neighbors, and the global shape structure. To alleviate inevitable global information loss caused by the rotation-invariant representations, we further introduce a region relation convolution to encode local and non-local information. We evaluate our method on multiple point cloud analysis tasks, including shape classification, part segmentation, and shape retrieval. Experimental results show that our method achieves consistent, and also the best performance, on inputs at arbitrary orientations, compared with the state-of-the-arts.

This paper presents a novel non-local part-aware deep neural network to denoise point clouds by exploring the inherent non-local self-similarity in 3D objects and scenes. Different from existing works that explore small local patches, we design the non-local learning unit (NLU) customized with a graph attention module to adaptively capture non-local semantically-related features over the entire point cloud. To enhance the denoising performance, we cascade a series of NLUs to progressively distill the noise features from the noisy inputs. Further, besides the conventional surface reconstruction loss, we formulate a semantic part loss to regularize the predictions towards the relevant parts and enable denoising in a part-aware manner. Lastly, we performed extensive experiments to evaluate our method, both quantitatively and qualitatively, and demonstrate its superiority over the state-of-the-arts on both synthetic and real-scanned noisy inputs.

We present PointAugment, a new auto-augmentation framework that automatically optimizes and augments point cloud samples to enrich the data diversity when we train a classification network. Different from existing auto-augmentation methods for 2D images, PointAugment is sample-aware and takes an adversarial learning strategy to jointly optimize an augmentor network and a classifier network, such that the augmentor can learn to produce augmented samples that best fit the classifier. Moreover, we formulate a learnable point augmentation function with a shape-wise transformation and a point-wise displacement, and carefully design loss functions to adopt the augmented samples based on the learning progress of the classifier. Extensive experiments also confirm PointAugment's effectiveness and robustness to improve the performance of various networks on shape classification and retrieval., Comment: Camera-Ready Version for CVPR 2020 (Oral); code is https://github.com/liruihui/PointAugment/

Point clouds acquired from range scans are often sparse, noisy, and non-uniform. This paper presents a new point cloud upsampling network called PU-GAN, which is formulated based on a generative adversarial network (GAN), to learn a rich variety of point distributions from the latent space and upsample points over patches on object surfaces. To realize a working GAN network, we construct an up-down-up expansion unit in the generator for upsampling point features with error feedback and self-correction, and formulate a self-attention unit to enhance the feature integration. Further, we design a compound loss with adversarial, uniform and reconstruction terms, to encourage the discriminator to learn more latent patterns and enhance the output point distribution uniformity. Qualitative and quantitative evaluations demonstrate the quality of our results over the state-of-the-arts in terms of distribution uniformity, proximity-to-surface, and 3D reconstruction quality., Comment: accepted by ICCV2019, project page at https://liruihui.github.io/publication/PU-GAN