Despite the unprecedented progress in the field of 3D generation, current systems still often fail to produce high-quality 3D assets that are visually appealing and geometrically and semantically consistent across multiple viewpoints. To effectively assess the quality of the generated 3D data, there is a need for a reliable 3D evaluation tool. Unfortunately, existing 3D evaluation metrics often overlook the geometric quality of generated assets or merely rely on black-box multimodal large language models for coarse assessment. In this paper, we introduce Eval3D, a fine-grained, interpretable evaluation tool that can faithfully evaluate the quality of generated 3D assets based on various distinct yet complementary criteria. Our key observation is that many desired properties of 3D generation, such as semantic and geometric consistency, can be effectively captured by measuring the consistency among various foundation models and tools. We thus leverage a diverse set of models and tools as probes to evaluate the inconsistency of generated 3D assets across different aspects. Compared to prior work, Eval3D provides pixel-wise measurement, enables accurate 3D spatial feedback, and aligns more closely with human judgments. We comprehensively evaluate existing 3D generation models using Eval3D and highlight the limitations and challenges of current models. Project page: http://eval3d.github.io.

Current vision systems typically assign fixed-length representations to images, regardless of the information content. This contrasts with human intelligence —and even large language models—which allocate varying representational capacities based on entropy, context and familiarity. Inspired by this, we propose an approach to learn variable-length token representations for 2D images. Our encoder-decoder architecture recursively processes 2D image tokens, distilling them into 1D latent tokens over multiple iterations of recurrent rollouts. Each iteration refines the 2D tokens, updates the existing 1D latent tokens, and adaptively increases representational capacity by adding new tokens. This enables compression of images into a variable number of tokens, ranging from 32 to 256. We validate our tokenizer using reconstruction loss and FID metrics, demonstrating that token count aligns with image entropy, familiarity and downstream task requirements. Recurrent token processing with increasing representational capacity in each iteration shows signs of token specialization, revealing potential for object / part discovery.

According to Algorithmic Information Theory (AIT), intelligent representations compress data into the shortest possible program while remaining predictive of its content—exhibiting low Kolmogorov Complexity (KC). In contrast, most visual representation learning systems assign fixed-length representations to all inputs, ignoring variations in complexity or familiarity. Recent adaptive tokenization methods address this by allocating variable-length representations but typically require test-time search over multiple hypotheses to identify the most predictive one. Inspired by KC principles, we propose a one-shot adaptive tokenizer, KARL, that predicts the appropriate number of tokens for an image in a single forward pass, halting once its approximate KC is reached. The token count serves as a proxy for the minimum description length. KARL performs comparably to recent adaptive tokenizers while operating in a one-pass manner. Additionally, we present a conceptual study showing a correlation between adaptive tokenization and core ideas from AIT. We demonstrate that adaptive tokenization not only aligns with KC but also reveals empirical signals approximating AIT concepts such as sophistication and logical depth. Finally, we analyze predicted image complexity and interestingness across axes such as structure vs. noise and in-distribution vs. out-of-distribution familiarity, highlighting alignment with human annotations.

What does learning to model relationships between strings teach Large Language Models (LLMs) about the visual world? We systematically evaluate LLMs' abilities to generate and recognize an assortment of visual concepts of increasing complexity and then demonstrate how a preliminary visual representation learning system can be trained using models of text. As language models lack the ability to consume or output visual information as pixels, we use code to represent images in our study. Although LLM-generated images do not look like natural images, results on image generation and the ability of models to correct these generated images indicate that precise modeling of strings can teach language models about numerous aspects of the visual world. Furthermore, experiments on self-supervised visual representation learning, utilizing images generated with text models, highlight the potential to train vision models capable of making semantic assessments of natural images using just LLMs.

The recent wave of large-scale text-to-image diffusion models has dramatically increased our text-based image generation abilities. These models can generate realistic images for a staggering variety of prompts and exhibit impressive compositional generalization abilities. Almost all use cases thus far have solely focused on sampling; however, diffusion models can also provide conditional density estimates, which are useful for tasks beyond image generation. In this paper, we show that the density estimates from large-scale text-to-image diffusion models like Stable Diffusion can be leveraged to perform zero-shot classification without any additional training. Our generative approach to classification, which we call Diffusion Classifier, attains strong results on a variety of benchmarks and outperforms alternative methods of extracting knowledge from diffusion models. Although a gap remains between generative and discriminative approaches on zero-shot recognition tasks, our diffusion-based approach has stronger multimodal compositional reasoning abilities than competing discriminative approaches. Finally, we use Diffusion Classifier to extract standard classifiers from class-conditional diffusion models trained on ImageNet. These models approach the performance of SOTA discriminative classifiers and exhibit strong "effective robustness" to distribution shift. Overall, our results are a step toward using generative over discriminative models for downstream tasks. Results and visualizations on our website: diffusion-classifier.github.io/

We present a new framework to learn dense 3D re-construction and correspondence from a single 2D image. The shape is represented implicitly as deformation over a category-level occupancy field and learned in an unsupervised manner from an unaligned image collection without using any 3D supervision. However, image collections usually contain large intra-category topological variation, e.g. images of different chair instances, posing a major challenge. Hence, prior methods are either restricted only to categories with no topological variation for estimating shape and correspondence or focus only on learning shape independently for each instance without any correspondence. To address this issue, we propose a topologically-aware deformation field that maps 3D points in object space to a higher-dimensional canonical space. Given a single image, we first implicitly deform a 3D point in the object space to a learned category-specific canonical space using the topologically-aware field and then learn the 3D shape in the canonical space. Both the canonical shape and deformation field are trained end-to-end using differentiable rendering via learned recurrent ray marcher. Our approach, dubbed TARS, achieves state-of-the-art reconstruction fidelity on several datasets: ShapeNet, Pascal3D+, CUB, and Pix3D chairs.

Reconstructing high-quality 3D objects from sparse, partial observations from a single view is of crucial importance for various applications in computer vision, robotics, and graphics. While recent neural implicit modeling methods show promising results on synthetic or dense data, they perform poorly on sparse and noisy real-world data. We discover that the limitations of a popular neural implicit model are due to lack of robust shape priors and lack of proper regularization. In this work, we demonstrate high-quality in-the-wild shape reconstruction using: (i) a deep encoder as a robust-initializer of the shape latent-code; (ii) regularized test-time optimization of the latent-code; (iii) a deep discriminator as a learned high-dimensional shape prior; (iv) a novel curriculum learning strategy that allows the model to learn shape priors on synthetic data and smoothly transfer them to sparse real world data. Our approach better captures the global structure, performs well on occluded and sparse observations, and registers well with the ground-truth shape. We demonstrate superior performance over state-of-the-art 3D object reconstruction methods on two real-world datasets.

Scalable sensor simulation is an important yet challenging open problem for safety-critical domains such as self-driving. Current works in image simulation either fail to be photorealistic or do not model the 3D environment and the dynamic objects within, losing high-level control and physical realism. In this paper, we present GeoSim, a geometry-aware image composition process which synthesizes novel urban driving scenarios by augmenting existing images with dynamic objects extracted from other scenes and rendered at novel poses. Towards this goal, we first build a diverse bank of 3D objects with both realistic geometry and appearance from sensor data. During simulation, we perform a novel geometry-aware simulation-by-composition procedure which 1) proposes plausible and realistic object placements into a given scene, 2) renders novel views of dynamic objects from the asset bank, and 3) composes and blends the rendered image segments. The resulting synthetic images are realistic, traffic-aware, and geometrically consistent, allowing our approach to scale to complex use cases. We demonstrate two such important applications: long-range realistic video simulation across multiple camera sensors, and synthetic data generation for data augmentation on downstream segmentation tasks. Please check https://tmux.top/publication/geosim/ for high-resolution video results.

Our goal is to significantly speed up the runtime of current state-of-the-art stereo algorithms to enable real-time inference. Towards this goal, we developed a differentiable PatchMatch module that allows us to discard most disparities without requiring full cost volume evaluation. We then exploit this representation to learn which range to prune for each pixel. By progressively reducing the search space and effectively propagating such information, we are able to efficiently compute the cost volume for high likelihood hypotheses and achieve savings in both memory and computation.Finally, an image guided refinement module is exploited to further improve the performance. Since all our components are differentiable, the full network can be trained end-to-end. Our experiments show that our method achieves competitive results on KITTI and SceneFlow datasets while running in real-time at 62ms.

Automatically generating the summary of a video is a challenging problem due to its subjective nature. Most of the previous works in the field consider the entire video to extract out the important frames. Unlike them, our paper presents MerryGoRoundNet, a supervised learning approach to solve this problem in an online fashion. We observe that to effectively summarize a video, one needs to take into account both the spatial and temporal relations between video frames. MerryGoRoundNet utilizes encoder-decoder style architecture and convolutional LSTM to establish spatiotemporal relationship and generates the summary on the fly, thereby being more efficient than non-autoregressive counterparts in terms of time and memory. In order to make summary more diverse and complete, we augment our network with unsupervised task of next frame prediction and a supervised task of scene start detection and propose a loss function that explicitly focuses on achieving the right balance between continuity and diversity in the produced summary. Ablation study performed affirms the architecture and learning objective of our approach. Evaluation of MerryGoRoundNet on different datasets demonstrates superior performance among online summarization approaches and competitive performance when compared with offline approaches as well.