Computer Vision – ECCV 2024

3D content creation has achieved significant progress in terms of both quality and speed. Although current feed-forward models can produce 3D objects in seconds, their resolution is constrained by the intensive computation required during training. In this paper, we introduce Large Multi-View Gaussian Model (LGM), a novel framework designed to generate high-resolution 3D models from text prompts or single-view images. Our key insights are two-fold: 1) 3D Representation: We propose multi-view Gaussian features as an efficient yet powerful representation, which can then be fused together for differentiable rendering. 2) 3D Backbone: We present an asymmetric U-Net as a high-throughput backbone operating on multi-view images, which can be produced from text or single-view image input by leveraging multi-view diffusion models. Extensive experiments demonstrate the high fidelity and efficiency of our approach. Notably, we maintain the fast speed to generate 3D objects within 5 s while boosting the training resolution to 512, thereby achieving high-resolution 3D content generation. Our project page is available at https://​me.​kiui.​moe/​lgm/​.

Multi-view crowd localization predicts the ground locations of all people in the scene. Typical methods usually estimate the crowd density maps on the ground plane first, and then obtain the crowd locations. However, existing methods’ performances are limited by the ambiguity of the density maps in crowded areas, where local peaks can be smoothed away. To mitigate the weakness of density map supervision, optimal transport-based point supervision methods have been proposed in the single-image crowd localization tasks, but have not been explored for multi-view crowd localization yet. Thus, in this paper, we propose a novel Mahalanobis distance-based multi-view optimal transport (M-MVOT) loss specifically designed for multi-view crowd localization. First, we replace the Euclidean-based transport cost with the Mahalanobis distance, which defines elliptical iso-contours in the cost function whose long-axis and short-axis directions are guided by the view ray direction. Second, the object-to-camera distance in each view is used to adjust the optimal transport cost of each location further, where the wrong predictions far away from the camera are more heavily penalized. Finally, we propose a strategy to consider all the input camera views in the model loss (M-MVOT) by computing the optimal transport cost for each ground-truth point based on its closest camera. Experiments demonstrate the advantage of the proposed method over density map-based or common Euclidean distance-based optimal transport loss on several multi-view crowd localization datasets.

sRGB images are now the predominant choice for pre-training visual models in computer vision research, owing to their ease of acquisition and efficient storage. Meanwhile, the advantage of RAW images lies in their rich physical information under variable real-world challenging lighting conditions. For computer vision tasks directly based on camera RAW data, most existing studies adopt methods of integrating image signal processor (ISP) with backend networks, yet often overlook the interaction capabilities between the ISP stages and subsequent networks. Drawing inspiration from ongoing adapter research in NLP and CV areas, we introduce RAW-Adapter, a novel approach aimed at adapting sRGB pre-trained models to camera RAW data. RAW-Adapter comprises input-level adapters that employ learnable ISP stages to adjust RAW inputs, as well as model-level adapters to build connections between ISP stages and subsequent high-level networks. Additionally, RAW-Adapter is a general framework that could be used in various computer vision frameworks. Abundant experiments under different lighting conditions have shown our algorithm’s state-of-the-art (SOTA) performance, demonstrating its effectiveness and efficiency across a range of real-world and synthetic datasets. Code is available at this url.

SLEDGE is the first generative simulator for vehicle motion planning trained on real-world driving logs. Its core component is a learned model that is able to generate agent bounding boxes and lane graphs. The model’s outputs serve as an initial state for rule-based traffic simulation. The unique properties of the entities to be generated for SLEDGE, such as their connectivity and variable count per scene, render the naive application of most modern generative models to this task non-trivial. Therefore, together with a systematic study of existing lane graph representations, we introduce a novel raster-to-vector autoencoder. It encodes agents and the lane graph into distinct channels in a rasterized latent map. This facilitates both lane-conditioned agent generation and combined generation of lanes and agents with a Diffusion Transformer. Using generated entities in SLEDGE enables greater control over the simulation, e.g. upsampling turns or increasing traffic density. Further, SLEDGE can support 500 m long routes, a capability not found in existing data-driven simulators like nuPlan. It presents new challenges for planning algorithms, evidenced by failure rates of over 40% for PDM, the winner of the 2023 nuPlan challenge, when tested on hard routes and dense traffic generated by our model. Compared to nuPlan, SLEDGE requires 500\(\times \) less storage to set up (<4 GB), making it a more accessible option and helping with democratizing future research in this field.

Object counting methods typically rely on manually annotated datasets. The cost of creating such datasets has restricted the versatility of these networks to count objects from specific classes (such as humans or penguins), and counting objects from diverse categories remains a challenge. The availability of robust text-to-image latent diffusion models (LDMs) raises the question of whether these models can be utilized to generate counting datasets. However, LDMs struggle to create images with an exact number of objects based solely on text prompts but they can be used to offer a dependable sorting signal by adding and removing objects within an image. Leveraging this data, we initially introduce an unsupervised sorting methodology to learn object-related features that are subsequently refined and anchored for counting purposes using counting data generated by LDMs. Further, we present a density classifier-guided method for dividing an image into patches containing objects that can be reliably counted. Consequently, we can generate counting data for any type of object and count them in an unsupervised manner. Our approach outperforms unsupervised and few-shot alternatives and is not restricted to specific object classes for which counting data is available. Code available at: github.​com/​adrian-dalessandro/​AFreeCA.

The field of neural rendering has witnessed significant progress with advancements in generative models and differentiable rendering techniques. Though 2D diffusion has achieved success, a unified 3D diffusion pipeline remains unsettled. This paper introduces a novel framework called LN3Diff to address this gap and enable fast, high-quality, and generic conditional 3D generation. Our approach harnesses a 3D-aware architecture and variational autoencoder (VAE) to encode the input image(s) into a structured, compact, and 3D latent space. The latent is decoded by a transformer-based decoder into a high-capacity 3D neural field. Through training a diffusion model on this 3D-aware latent space, our method achieves superior performance on Objaverse, ShapeNet and FFHQ for conditional 3D generation. Moreover, it surpasses existing 3D diffusion methods in terms of inference speed, requiring no per-instance optimization. Video demos can be found on our project webpage: https://​nirvanalan.​github.​io/​projects/​ln3diff.

Camera-based 3D semantic scene completion (SSC) is pivotal for predicting complicated 3D layouts with limited 2D image observations. The existing mainstream solutions generally leverage temporal information by roughly stacking history frames to supplement the current frame, such straightforward temporal modeling inevitably diminishes valid clues and increases learning difficulty. To address this problem, we present HTCL, a novel Hierarchical Temporal Context Learning paradigm for improving camera-based semantic scene completion. The primary innovation of this work involves decomposing temporal context learning into two hierarchical steps: (a) cross-frame affinity measurement and (b) affinity-based dynamic refinement. Firstly, to separate critical relevant context from redundant information, we introduce the pattern affinity with scale-aware isolation and multiple independent learners for fine-grained contextual correspondence modeling. Subsequently, to dynamically compensate for incomplete observations, we adaptively refine the feature sampling locations based on initially identified locations with high affinity and their neighboring relevant regions. Our method ranks \(1^{st}\) on the SemanticKITTI benchmark and even surpasses LiDAR-based methods in terms of mIoU on the OpenOccupancy benchmark. Our code is available on https://​github.​com/​Arlo0o/​HTCL.

Point cloud registration is a foundational task for 3D alignment and reconstruction applications. While both traditional and learning-based registration approaches have succeeded, leveraging the intrinsic symmetry of point cloud data, including rotation equivariance, has received insufficient attention. This prohibits the model from learning effectively, resulting in a requirement for more training data and increased model complexity. To address these challenges, we propose a graph neural network model embedded with a local Spherical Euclidean 3D equivariance property through \(\textbf{SE}\)(3) message passing based propagation. Our model is composed mainly of a descriptor module, equivariant graph layers, match similarity, and the final regression layers. Such modular design enables us to utilize sparsely sampled input points and initialize the descriptor by self-trained or pre-trained geometric feature descriptors easily. Experiments conducted on the 3DMatch and KITTI datasets exhibit the compelling and robust performance of our model compared to state-of-the-art approaches, while the model complexity remains relatively low at the same time.

Recent advances in learning multi-modal representation have witnessed the success in biomedical domains. While established techniques enable handling multi-modal information, the challenges are posed when extended to various clinical modalities and practical modality-missing setting due to the inherent modality gaps. To tackle these, we propose an innovative Modality-prompted Heterogeneous Graph for Omni-modal Learning (GTP-4o), which embeds the numerous disparate clinical modalities into a unified representation, completes the deficient embedding of missing modality and reformulates the cross-modal learning with a graph-based aggregation. Specially, we establish a heterogeneous graph embedding to explicitly capture the diverse semantic properties on both the modality-specific features (nodes) and the cross-modal relations (edges). Then, we design a modality-prompted completion that enables completing the inadequate graph representation of missing modality through a graph prompting mechanism, which generates hallucination graphic topologies to steer the missing embedding towards the intact representation. Through the completed graph, we meticulously develop a knowledge-guided hierarchical cross-modal aggregation consisting of a global meta-path neighbouring to uncover the potential heterogeneous neighbors along the pathways driven by domain knowledge, and a local multi-relation aggregation module for the comprehensive cross-modal interaction across various heterogeneous relations. We assess the efficacy of our methodology on rigorous benchmarking experiments against prior state-of-the-arts. In a nutshell, GTP-4o presents an initial foray into the intriguing realm of embedding, relating and perceiving the heterogeneous patterns from various clinical modalities holistically via a graph theory. Project page: https://​gtp4-o.​github.​io/​.

We tackle the problem of Continual Category Discovery (CCD), which aims to automatically discover novel categories in a continuous stream of unlabeled data while mitigating the challenge of catastrophic forgetting—an open problem that persists even in conventional, fully supervised continual learning. To address this challenge, we propose PromptCCD, a simple yet effective framework that utilizes a Gaussian Mixture Model (GMM) as a prompting method for CCD. At the core of PromptCCD lies the Gaussian Mixture Prompting (GMP) module, which acts as a dynamic pool that updates over time to facilitate representation learning and prevent forgetting during category discovery. Moreover, GMP enables on-the-fly estimation of category numbers, allowing PromptCCD to discover categories in unlabeled data without prior knowledge of the category numbers. We extend the standard evaluation metric for Generalized Category Discovery (GCD) to CCD and benchmark state-of-the-art methods on diverse public datasets. PromptCCD significantly outperforms existing methods, demonstrating its effectiveness. Project page: https://​visual-ai.​github.​io/​promptccd.

“Sapiens-pertaining to, or resembling modern humans.”

This paper introduces an approach to linearly controllable generative adversarial networks (LC-GAN) driven by unsupervised learning. Departing from traditional methods relying on supervision signals or post-processing for latent feature disentanglement, our proposed technique enables unsupervised learning using only image data through contrastive feature categorization and spectral regularization. In our framework, the discriminator constructs geometry- and appearance-related feature spaces using a combination of image augmentation and contrastive representation learning. Leveraging these feature spaces, the generator autonomously categorizes input latent codes into geometry- and appearance-related features. Subsequently, the categorized features undergo projection into a subspace via our proposed spectral regularization, with each component controlling a distinct aspect of the generated image. Beyond providing fine-grained control over the generative model, our approach achieves state-of-the-art image generation quality on benchmark datasets, including FFHQ, CelebA-HQ, and AFHQ-V2.

We present TeSMo, a text-controlled scene-aware motion generation method based on denoising diffusion models. Previous text-to-motion methods focus on characters in isolation without considering scenes due to the limited availability of datasets that include motion, text descriptions, and interactive scenes. Our approach begins with pre-training a scene-agnostic text-to-motion diffusion model, emphasizing goal-reaching constraints on large-scale motion-capture datasets. We then enhance this model with a scene-aware component, fine-tuned using data augmented with detailed scene information, including ground plane and object shapes. To facilitate training, we embed annotated navigation and interaction motions within scenes. The proposed method produces realistic and diverse human-object interactions, such as navigation and sitting, in different scenes with various object shapes, orientations, initial body positions, and poses. Extensive experiments demonstrate that our approach surpasses prior techniques in terms of the plausibility of human-scene interactions and the realism and variety of the generated motions. Code and data are available at https://​research.​nvidia.​com/​labs/​toronto-ai/​tesmo.

Discriminative models for object classification typically learn image-based representations that do not capture the compositional and 3D nature of objects. In this work, we show that explicitly integrating 3D compositional object representations into deep networks for image classification leads to a largely enhanced generalization in out-of-distribution scenarios. In particular, we introduce a novel architecture, referred to as NOVUM, that consists of a feature extractor and a neural object volume for every target object class. Each neural object volume is a composition of 3D Gaussians that emit feature vectors. This compositional object representation allows for a highly robust and fast estimation of the object class by independently matching the features of the 3D Gaussians of each category to features extracted from an input image. Additionally, the object pose can be estimated via inverse rendering of the corresponding neural object volume. To enable the classification of objects, the neural features at each 3D Gaussian are trained discriminatively to be distinct from (i) the features of 3D Gaussians in other categories, (ii) features of other 3D Gaussians of the same object, and (iii) the background features. Our experiments show that NOVUM offers intriguing advantages over standard architectures due to the 3D compositional structure of the object representation, namely: (1) An exceptional robustness across a spectrum of real-world and synthetic out-of-distribution shifts and (2) an enhanced human interpretability compared to standard models, all while maintaining real-time inference and a competitive accuracy on in-distribution data. Code and model can be found at

Collaborative perception has received widespread attention recently since it enhances the perception ability of autonomous vehicles via inter-agent information sharing. However, the performance of existing systems is hindered by the unavoidable collaboration noises, which induce feature-level spatial misalignment over the collaborator-shared information. In this paper, we propose a model-agnostic and lightweight plugin to mitigate the feature-level misalignment issue, called dynamic feature alignment (NEAT). The merits of the NEAT plugin are threefold. First, we introduce an importance-guided query proposal to predict potential foreground regions with space-channel semantics and exclude environmental redundancies. On this basis, a deformable feature alignment is presented to explicitly align the collaborator-shared features through query-aware spatial associations, aggregating multi-grained visual clues with corrective mismatch properties. Ultimately, we perform a region cross-attention reinforcement to facilitate aligned representation diffusion and achieve global feature semantic enhancement. NEAT can be readily inserted into existing collaborative perception procedures and significantly improves the robustness of vanilla baselines against pose errors and transmission delay. Extensive experiments on four collaborative 3D object detection datasets under noisy settings confirm that NEAT provides consistent gains for most methods with distinct structures.

Generating human-object interactions (HOIs) is critical with the tremendous advances of digital avatars. Existing datasets are typically limited to humans interacting with a single object while neglecting the ubiquitous manipulation of multiple objects. Thus, we propose HIMO, a large-scale MoCap dataset of full-body human interacting with multiple objects, containing 3.3K 4D HOI sequences and 4.08M 3D HOI frames. We also annotate HIMO with detailed textual descriptions and temporal segments, benchmarking two novel tasks of HOI synthesis conditioned on either the whole text prompt or the segmented text prompts as fine-grained timeline control. To address these novel tasks, we propose a dual-branch conditional diffusion model with a mutual interaction module for HOI synthesis. Besides, an auto-regressive generation pipeline is also designed to obtain smooth transitions between HOI segments. Experimental results demonstrate the generalization ability to unseen object geometries and temporal compositions. Our data, codes, and models will be publicly available for research purposes.

Implicit representation of an image can map arbitrary coordinates in the continuous domain to their corresponding color values, presenting a powerful capability for image reconstruction. Nevertheless, existing implicit representation approaches only focus on building continuous appearance mapping, ignoring the continuities of the semantic information across pixels. Consequently, achieving the desired reconstruction results becomes challenging when the semantic information within input image is corrupted, such as when a large region is missing. To address the issue, we suggest learning semantic-aware implicit representation (SAIR), that is, we make the implicit representation of each pixel rely on both its appearance and semantic information (e.g., which object does the pixel belong to). To this end, we propose a framework with two modules: (1) a semantic implicit representation (SIR) for a corrupted image. Given an arbitrary coordinate in the continuous domain, we can obtain its respective text-aligned embedding indicating the object the pixel belongs. (2) an appearance implicit representation (AIR) based on the SIR. Given an arbitrary coordinate in the continuous domain, we can reconstruct its color whether or not the pixel is missed in the input. We validate the novel semantic-aware implicit representation method on the image inpainting task, and the extensive experiments demonstrate that our method surpasses state-of-the-art approaches by a significant margin.

How to effectively explore spatial-temporal features is important for video colorization. Instead of stacking multiple frames along the temporal dimension or recurrently propagating estimated features that will accumulate errors or cannot explore information from far-apart frames, we develop a memory-based feature propagation module that can establish reliable connections with features from far-apart frames and alleviate the influence of inaccurately estimated features. To extract better features from each frame for the above-mentioned feature propagation, we explore the features from large-pretrained visual models to guide the feature estimation of each frame so that the estimated features can model complex scenarios. In addition, we note that adjacent frames usually contain similar contents. To explore this property for better spatial and temporal feature utilization, we develop a local attention module to aggregate the features from adjacent frames in a spatial-temporal neighborhood. We formulate our memory-based feature propagation module, large-pretrained visual model guided feature estimation module, and local attention module into an end-to-end trainable network (named ColorMNet) and show that it performs favorably against state-of-the-art methods on both the benchmark datasets and real-world scenarios. Our source codes and pre-trained models are available at: https://​github.​com/​yyang181/​colormnet.

Deep learning faces a formidable challenge when handling noisy labels, as models tend to overfit samples affected by label noise. This challenge is further compounded by the presence of instance-dependent noise (IDN), a realistic form of label noise arising from ambiguous sample information. To address IDN, Label Noise Learning (LNL) incorporates a sample selection stage to differentiate clean and noisy-label samples. This stage uses an arbitrary criterion and a pre-defined curriculum that initially selects most samples as noisy and gradually decreases this selection rate during training. Such curriculum is sub-optimal since it does not consider the actual label noise rate in the training set. This paper addresses this issue with a new noise-rate estimation method that is easily integrated with most state-of-the-art (SOTA) LNL methods to produce a more effective curriculum. Synthetic and real-world benchmarks’ results demonstrate that integrating our approach with SOTA LNL methods improves accuracy in most cases. (Code is available at https://​github.​com/​arpit2412/​NoiseRateLearnin​g. Supported by the Engineering and Physical Sciences Research Council (EPSRC) through grant EP/Y018036/1 and the Australian Research Council (ARC) through grant FT190100525.)

In this paper, we retarget video stitching to an emerging issue, named warping shake, when extending image stitching to video stitching. It unveils the temporal instability of warped content in non-overlapping regions, despite image stitching having endeavored to preserve the natural structures. Therefore, in most cases, even if the input videos to be stitched are stable, the stitched video will inevitably cause undesired warping shakes and affect the visual experience. To eliminate the shakes, we propose StabStitch to simultaneously realize video stitching and video stabilization in a unified unsupervised learning framework. Starting from the camera paths in video stabilization, we first derive the expression of stitching trajectories in video stitching by elaborately integrating spatial and temporal warps. Then a warp smoothing model is presented to optimize them with a comprehensive consideration regarding content alignment, trajectory smoothness, spatial consistency, and online collaboration. To establish an evaluation benchmark and train the learning framework, we build a video stitching dataset with a rich diversity in camera motions and scenes. Compared with existing stitching solutions, StabStitch exhibits significant superiority in scene robustness and inference speed in addition to stitching and stabilization performance, contributing to a robust and real-time online video stitching system. The codes and dataset are available at https://​github.​com/​nie-lang/​StabStitch.

Most Large Vision-Language Models (LVLMs) enjoy the same vision vocabulary, i.e., CLIP, for common vision tasks. However, for some special task that needs dense and fine-grained perception, the CLIP-style vocabulary may encounter low efficiency in tokenizing corresponding vision knowledge and even suffer out-of-vocabulary problems. Accordingly, we propose Vary, an efficient and productive method to scale up the Vision vocabulary of LVLMs. The procedures of Vary are naturally divided into two folds: the generation and integration of a new vision vocabulary. In the first phase, we devise a vocabulary network along with a tiny decoder-only transformer to compress rich vision signals. Next, we scale up the vanilla vision vocabulary by merging the new with the original one (CLIP), enabling the LVLMs to garner new features effectively. We present frameworks with two sizes: Vary-base (7B) and Vary-toy (1.8B), both of which enjoy excellent fine-grained perception performance while maintaining great general ability.

Humans can foresee the future based on present observations, a skill we term as foresight minds. However, this capability remains under-explored within existing MLLMs, hindering their capacity to understand intentions behind subjects. To address this, we integrate the future modeling into MLLMs. By utilizing the trajectory, a highly structured representation, as a learning objective, we aim to equip the model to understand spatiotemporal dynamics. Inspired by the learning paradigm of LLMs, we first propose Foresight Pre-Training (FPT) that jointly learns various tasks centered on trajectories, enabling MLLMs to predict entire trajectories from a given initial observation. Then, we propose Foresight Instruction-Tuning (FIT) that requires MLLMs to reason about potential future events based on predicted trajectories. Aided by FPT and FIT, we build an unified MLLM named Merlin that supports complex future reasoning. Experiments show Merlin’s foresight minds with impressive performance on both future reasoning and visual comprehension tasks. Project page: https://​ahnsun.​github.​io/​merlin.

We propose ViC-MAE, a model that combines both Masked AutoEncoders (MAE) and contrastive learning. ViC-MAE is trained using a global representation obtained by pooling the local features learned under an MAE reconstruction loss and using this representation under a contrastive objective across images and video frames. We show that visual representations learned under ViC-MAE generalize well to video and image classification tasks. Particularly, ViC-MAE obtains state-of-the-art transfer learning performance from video to images on Imagenet-1k compared to the recently proposed OmniMAE by achieving a top-1 accuracy of 86% (+1.3% absolute improvement) when trained on the same data and 87.1% (+2.4% absolute improvement) when training on extra data. At the same time, ViC-MAE outperforms most other methods on video benchmarks by obtaining 75.9% top-1 accuracy on the challenging Something something-v2 video benchmark. When training on videos and images from diverse datasets, our method maintains a balanced transfer-learning performance between video and image classification benchmarks, coming only as a close second to the best-supervised method.

Surgical scene perception via videos is critical for advancing robotic surgery, telesurgery, and AI-assisted surgery, particularly in ophthalmology. However, the scarcity of diverse and richly annotated video datasets has hindered the development of intelligent systems for surgical workflow analysis. Existing datasets face challenges such as small scale, lack of diversity in surgery and phase categories, and absence of time-localized annotations. These limitations impede action understanding and model generalization validation in complex and diverse real-world surgical scenarios. To address this gap, we introduce OphNet, a large-scale, expert-annotated video benchmark for ophthalmic surgical workflow understanding. OphNet features: 1) A diverse collection of 2,278 surgical videos spanning 66 types of cataract, glaucoma, and corneal surgeries, with detailed annotations for 102 unique surgical phases and 150 fine-grained operations. 2) Sequential and hierarchical annotations for each surgery, phase, and operation, enabling comprehensive understanding and improved interpretability. 3) Time-localized annotations, facilitating temporal localization and prediction tasks within surgical workflows. With approximately 285 h of surgical videos, OphNet is about 20 times larger than the largest existing surgical workflow analysis benchmark. Code and dataset are available at: https://​minghu0830.​github.​io/​OphNet-benchmark/​.