Igor Vasiljevic

Vitor Guizilini, Igor Vasiljevic, Dian Chen et al.

Monocular depth estimation is scale-ambiguous, and thus requires scale supervision to produce metric predictions. Even so, the resulting models will be geometry-specific, with learned scales that cannot be directly transferred across domains. Because of that, recent works focus instead on relative depth, eschewing scale in favor of improved up-to-scale zero-shot transfer. In this work we introduce ZeroDepth, a novel monocular depth estimation framework capable of predicting metric scale for arbitrary test images from different domains and camera parameters. This is achieved by (i) the use of input-level geometric embeddings that enable the network to learn a scale prior over objects; and (ii) decoupling the encoder and decoder stages, via a variational latent representation that is conditioned on single frame information. We evaluated ZeroDepth targeting both outdoor (KITTI, DDAD, nuScenes) and indoor (NYUv2) benchmarks, and achieved a new state-of-the-art in both settings using the same pre-trained model, outperforming methods that train on in-domain data and require test-time scaling to produce metric estimates.

Vitor Guizilini, Igor Vasiljevic, Jiading Fang et al.

Modern 3D computer vision leverages learning to boost geometric reasoning, mapping image data to classical structures such as cost volumes or epipolar constraints to improve matching. These architectures are specialized according to the particular problem, and thus require significant task-specific tuning, often leading to poor domain generalization performance. Recently, generalist Transformer architectures have achieved impressive results in tasks such as optical flow and depth estimation by encoding geometric priors as inputs rather than as enforced constraints. In this paper, we extend this idea and propose to learn an implicit, multi-view consistent scene representation, introducing a series of 3D data augmentation techniques as a geometric inductive prior to increase view diversity. We also show that introducing view synthesis as an auxiliary task further improves depth estimation. Our Depth Field Networks (DeFiNe) achieve state-of-the-art results in stereo and video depth estimation without explicit geometric constraints, and improve on zero-shot domain generalization by a wide margin.

Arkadeep Narayan Chaudhury, Igor Vasiljevic, Sergey Zakharov et al.

Synthesizing accurate geometry and photo-realistic appearance of small scenes is an active area of research with compelling use cases in gaming, virtual reality, robotic-manipulation, autonomous driving, convenient product capture, and consumer-level photography. When applying scene geometry and appearance estimation techniques to robotics, we found that the narrow cone of possible viewpoints due to the limited range of robot motion and scene clutter caused current estimation techniques to produce poor quality estimates or even fail. On the other hand, in robotic applications, dense metric depth can often be measured directly using stereo and illumination can be controlled. Depth can provide a good initial estimate of the object geometry to improve reconstruction, while multi-illumination images can facilitate relighting. In this work we demonstrate a method to incorporate dense metric depth into the training of neural 3D representations and address an artifact observed while jointly refining geometry and appearance by disambiguating between texture and geometry edges. We also discuss a multi-flash stereo camera system developed to capture the necessary data for our pipeline and show results on relighting and view synthesis with a few training views.

Basile Van Hoorick, Dian Chen, Shun Iwase et al. · gatech

Modern generative video models excel at producing convincing, high-quality outputs, but struggle to maintain multi-view and spatiotemporal consistency in highly dynamic real-world environments. In this work, we introduce \textbf{AnyView}, a diffusion-based video generation framework for \emph{dynamic view synthesis} with minimal inductive biases or geometric assumptions. We leverage multiple data sources with various levels of supervision, including monocular (2D), multi-view static (3D) and multi-view dynamic (4D) datasets, to train a generalist spatiotemporal implicit representation capable of producing zero-shot novel videos from arbitrary camera locations and trajectories. We evaluate AnyView on standard benchmarks, showing competitive results with the current state of the art, and propose \textbf{AnyViewBench}, a challenging new benchmark tailored towards \emph{extreme} dynamic view synthesis in diverse real-world scenarios. In this more dramatic setting, we find that most baselines drastically degrade in performance, as they require significant overlap between viewpoints, while AnyView maintains the ability to produce realistic, plausible, and spatiotemporally consistent videos when prompted from \emph{any} viewpoint. Results, data, code, and models can be viewed at: https://tri-ml.github.io/AnyView/

Takayuki Kanai, Igor Vasiljevic, Vitor Guizilini et al.

Autonomous vehicles and robots need to operate over a wide variety of scenarios in order to complete tasks efficiently and safely. Multi-camera self-supervised monocular depth estimation from videos is a promising way to reason about the environment, as it generates metrically scaled geometric predictions from visual data without requiring additional sensors. However, most works assume well-calibrated extrinsics to fully leverage this multi-camera setup, even though accurate and efficient calibration is still a challenging problem. In this work, we introduce a novel method for extrinsic calibration that builds upon the principles of self-supervised monocular depth and ego-motion learning. Our proposed curriculum learning strategy uses monocular depth and pose estimators with velocity supervision to estimate extrinsics, and then jointly learns extrinsic calibration along with depth and pose for a set of overlapping cameras rigidly attached to a moving vehicle. Experiments on a benchmark multi-camera dataset (DDAD) demonstrate that our method enables self-calibration in various scenes robustly and efficiently compared to a traditional vision-based pose estimation pipeline. Furthermore, we demonstrate the benefits of extrinsics self-calibration as a way to improve depth prediction via joint optimization.

Vitor Guizilini, Igor Vasiljevic, Jiading Fang et al.

Differentiable volumetric rendering is a powerful paradigm for 3D reconstruction and novel view synthesis. However, standard volume rendering approaches struggle with degenerate geometries in the case of limited viewpoint diversity, a common scenario in robotics applications. In this work, we propose to use the multi-view photometric objective from the self-supervised depth estimation literature as a geometric regularizer for volumetric rendering, significantly improving novel view synthesis without requiring additional information. Building upon this insight, we explore the explicit modeling of scene geometry using a generalist Transformer, jointly learning a radiance field as well as depth and light fields with a set of shared latent codes. We demonstrate that sharing geometric information across tasks is mutually beneficial, leading to improvements over single-task learning without an increase in network complexity. Our DeLiRa architecture achieves state-of-the-art results on the ScanNet benchmark, enabling high quality volumetric rendering as well as real-time novel view and depth synthesis in the limited viewpoint diversity setting.

Samir Yitzhak Gadre, Georgios Smyrnis, Vaishaal Shankar et al. · allen-ai, cmu

Scaling laws are useful guides for derisking expensive training runs, as they predict performance of large models using cheaper, small-scale experiments. However, there remain gaps between current scaling studies and how language models are ultimately trained and evaluated. For instance, scaling is usually studied in the compute-optimal training regime (i.e., "Chinchilla optimal" regime). In contrast, models are often over-trained to reduce inference costs. Moreover, scaling laws mostly predict loss on next-token prediction, but models are usually compared on downstream task performance. To address both shortcomings, we create a testbed of 104 models with 0.011B to 6.9B parameters trained with various numbers of tokens on three data distributions. First, we fit scaling laws that extrapolate in both the amount of over-training and the number of model parameters. This enables us to predict the validation loss of a 1.4B parameter, 900B token run (i.e., 32$\times$ over-trained) and a 6.9B parameter, 138B token run (i.e., a compute-optimal run)$\unicode{x2014}$each from experiments that take 300$\times$ less compute. Second, we relate the perplexity of a language model to its downstream task performance by proposing a power law. We use this law to predict top-1 error averaged over downstream tasks for the two aforementioned models, using experiments that take 20$\times$ less compute. Our experiments are available at https://github.com/mlfoundations/scaling.

Jean Mercat, Igor Vasiljevic, Sedrick Keh et al. · cmu

Linear transformers have emerged as a subquadratic-time alternative to softmax attention and have garnered significant interest due to their fixed-size recurrent state that lowers inference cost. However, their original formulation suffers from poor scaling and underperforms compute-matched transformers. Recent linear models such as RWKV and Mamba have attempted to address these shortcomings by proposing novel time-mixing and gating architectures, but pre-training large language models requires significant data and compute investments. Thus, the search for subquadratic architectures is limited by the availability of compute and quality pre-training datasets. As a cost-effective alternative to pre-training linear transformers, we propose Scalable UPtraining for Recurrent Attention (SUPRA). We present a method to uptrain existing large pre-trained transformers into Recurrent Neural Networks (RNNs) with a modest compute budget. This allows us to leverage the strong pre-training data and performance of existing transformer LLMs, while requiring 5% of the training cost. We find that our linearization technique leads to competitive performance on standard benchmarks, but we identify persistent in-context learning and long-context modeling shortfalls for even the largest linear models. Our code and models can be found at https://github.com/TRI-ML/linear_open_lm.

Igor Vasiljevic

Modern computer vision has moved beyond the domain of internet photo collections and into the physical world, guiding camera-equipped robots and autonomous cars through unstructured environments. To enable these embodied agents to interact with real-world objects, cameras are increasingly being used as depth sensors, reconstructing the environment for a variety of downstream reasoning tasks. Machine-learning-aided depth perception, or depth estimation, predicts for each pixel in an image the distance to the imaged scene point. While impressive strides have been made in depth estimation, significant challenges remain: (1) ground truth depth labels are difficult and expensive to collect at scale, (2) camera information is typically assumed to be known, but is often unreliable and (3) restrictive camera assumptions are common, even though a great variety of camera types and lenses are used in practice. In this thesis, we focus on relaxing these assumptions, and describe contributions toward the ultimate goal of turning cameras into truly generic depth sensors.

Jiading Fang, Xiangshan Tan, Shengjie Lin et al.

If robots are to work effectively alongside people, they must be able to interpret natural language references to objects in their 3D environment. Understanding 3D referring expressions is challenging -- it requires the ability to both parse the 3D structure of the scene and correctly ground free-form language in the presence of distraction and clutter. We introduce Transcrib3D, an approach that brings together 3D detection methods and the emergent reasoning capabilities of large language models (LLMs). Transcrib3D uses text as the unifying medium, which allows us to sidestep the need to learn shared representations connecting multi-modal inputs, which would require massive amounts of annotated 3D data. As a demonstration of its effectiveness, Transcrib3D achieves state-of-the-art results on 3D reference resolution benchmarks, with a great leap in performance from previous multi-modality baselines. To improve upon zero-shot performance and facilitate local deployment on edge computers and robots, we propose self-correction for fine-tuning that trains smaller models, resulting in performance close to that of large models. We show that our method enables a real robot to perform pick-and-place tasks given queries that contain challenging referring expressions. Project site is at https://ripl.github.io/Transcrib3D.

Jiahao Li, Haochen Wang, Muhammad Zubair Irshad et al. · gatech

We propose FastMap, a new global structure from motion method focused on speed and simplicity. Previous methods like COLMAP and GLOMAP are able to estimate high-precision camera poses, but suffer from poor scalability when the number of matched keypoint pairs becomes large, mainly due to the time-consuming process of second-order Gauss-Newton optimization. Instead, we design our method solely based on first-order optimizers. To obtain maximal speedup, we identify and eliminate two key performance bottlenecks: computational complexity and the kernel implementation of each optimization step. Through extensive experiments, we show that FastMap is up to 10 times faster than COLMAP and GLOMAP with GPU acceleration and achieves comparable pose accuracy.

Cristobal Eyzaguirre, Igor Vasiljevic, Achal Dave et al. · salesforce, stanford

We propose a data-driven approach to analyzing query complexity in Video Question Answering (VideoQA). Previous efforts in benchmark design have relied on human expertise to design challenging questions, yet we experimentally show that humans struggle to predict which questions are difficult for machine learning models. Our automatic approach leverages recent advances in code generation for visual question answering, using the complexity of generated code as a proxy for question difficulty. We demonstrate that this measure correlates significantly better with model performance than human estimates. To operationalize this insight, we propose an algorithm for estimating question complexity from code. It identifies fine-grained primitives that correlate with the hardest questions for any given set of models, making it easy to scale to new approaches in the future. Finally, to further illustrate the utility of our method, we extend it to automatically generate complex questions, constructing a new benchmark that is 1.9 times harder than the popular NExT-QA.

Sedrick Keh, Jean Mercat, Samir Yitzhak Gadre et al. · cmu

Pre-trained LLMs that are further trained with image data perform well on vision-language tasks. While adding images during a second training phase effectively unlocks this capability, it is unclear how much of a gain or loss this two-step pipeline gives over VLMs which integrate images earlier into the training process. To investigate this, we train models spanning various datasets, scales, image-text ratios, and amount of pre-training done before introducing vision tokens. We then fine-tune these models and evaluate their downstream performance on a suite of vision-language and text-only tasks. We find that pre-training with a mixture of image and text data allows models to perform better on vision-language tasks while maintaining strong performance on text-only evaluations. On an average of 6 diverse tasks, we find that for a 1B model, introducing visual tokens 80% of the way through pre-training results in a 2% average improvement over introducing visual tokens to a fully pre-trained model.

Jeffrey Li, Alex Fang, Georgios Smyrnis et al.

We introduce DataComp for Language Models (DCLM), a testbed for controlled dataset experiments with the goal of improving language models. As part of DCLM, we provide a standardized corpus of 240T tokens extracted from Common Crawl, effective pretraining recipes based on the OpenLM framework, and a broad suite of 53 downstream evaluations. Participants in the DCLM benchmark can experiment with data curation strategies such as deduplication, filtering, and data mixing at model scales ranging from 412M to 7B parameters. As a baseline for DCLM, we conduct extensive experiments and find that model-based filtering is key to assembling a high-quality training set. The resulting dataset, DCLM-Baseline enables training a 7B parameter language model from scratch to 64% 5-shot accuracy on MMLU with 2.6T training tokens. Compared to MAP-Neo, the previous state-of-the-art in open-data language models, DCLM-Baseline represents a 6.6 percentage point improvement on MMLU while being trained with 40% less compute. Our baseline model is also comparable to Mistral-7B-v0.3 and Llama 3 8B on MMLU (63% & 66%), and performs similarly on an average of 53 natural language understanding tasks while being trained with 6.6x less compute than Llama 3 8B. Our results highlight the importance of dataset design for training language models and offer a starting point for further research on data curation.

Takayuki Kanai, Igor Vasiljevic, Vitor Guizilini et al.

Monocular visual odometry is a key technology in various autonomous systems. Traditional feature-based methods suffer from failures due to poor lighting, insufficient texture, and large motions. In contrast, recent learning-based dense SLAM methods exploit iterative dense bundle adjustment to address such failure cases, and achieve robust and accurate localization in a wide variety of real environments, without depending on domain-specific supervision. However, despite its potential, the methods still struggle with scenarios involving large motion and object dynamics. In this study, we diagnose key weaknesses in a popular learning-based dense SLAM model (DROID-SLAM) by analyzing major failure cases on outdoor benchmarks and exposing various shortcomings of its optimization process. We then propose the use of self-supervised priors leveraging a frozen large-scale pre-trained monocular depth estimator to initialize the dense bundle adjustment process, leading to robust visual odometry without the need to fine-tune the SLAM backbone. Despite its simplicity, the proposed method demonstrates significant improvements on KITTI odometry, as well as the challenging DDAD benchmark.

Jiading Fang, Shengjie Lin, Igor Vasiljevic et al.

A practical benefit of implicit visual representations like Neural Radiance Fields (NeRFs) is their memory efficiency: large scenes can be efficiently stored and shared as small neural nets instead of collections of images. However, operating on these implicit visual data structures requires extending classical image-based vision techniques (e.g., registration, blending) from image sets to neural fields. Towards this goal, we propose NeRFuser, a novel architecture for NeRF registration and blending that assumes only access to pre-generated NeRFs, and not the potentially large sets of images used to generate them. We propose registration from re-rendering, a technique to infer the transformation between NeRFs based on images synthesized from individual NeRFs. For blending, we propose sample-based inverse distance weighting to blend visual information at the ray-sample level. We evaluate NeRFuser on public benchmarks and a self-collected object-centric indoor dataset, showing the robustness of our method, including to views that are challenging to render from the individual source NeRFs.

Jiading Fang, Igor Vasiljevic, Vitor Guizilini et al.

Camera calibration is integral to robotics and computer vision algorithms that seek to infer geometric properties of the scene from visual input streams. In practice, calibration is a laborious procedure requiring specialized data collection and careful tuning. This process must be repeated whenever the parameters of the camera change, which can be a frequent occurrence for mobile robots and autonomous vehicles. In contrast, self-supervised depth and ego-motion estimation approaches can bypass explicit calibration by inferring per-frame projection models that optimize a view synthesis objective. In this paper, we extend this approach to explicitly calibrate a wide range of cameras from raw videos in the wild. We propose a learning algorithm to regress per-sequence calibration parameters using an efficient family of general camera models. Our procedure achieves self-calibration results with sub-pixel reprojection error, outperforming other learning-based methods. We validate our approach on a wide variety of camera geometries, including perspective, fisheye, and catadioptric. Finally, we show that our approach leads to improvements in the downstream task of depth estimation, achieving state-of-the-art results on the EuRoC dataset with greater computational efficiency than contemporary methods.

Vitor Guizilini, Igor Vasiljevic, Rares Ambrus et al.

Self-supervised monocular depth and ego-motion estimation is a promising approach to replace or supplement expensive depth sensors such as LiDAR for robotics applications like autonomous driving. However, most research in this area focuses on a single monocular camera or stereo pairs that cover only a fraction of the scene around the vehicle. In this work, we extend monocular self-supervised depth and ego-motion estimation to large-baseline multi-camera rigs. Using generalized spatio-temporal contexts, pose consistency constraints, and carefully designed photometric loss masking, we learn a single network generating dense, consistent, and scale-aware point clouds that cover the same full surround 360 degree field of view as a typical LiDAR scanner. We also propose a new scale-consistent evaluation metric more suitable to multi-camera settings. Experiments on two challenging benchmarks illustrate the benefits of our approach over strong baselines.

Igor Vasiljevic, Vitor Guizilini, Rares Ambrus et al.

Self-supervised learning has emerged as a powerful tool for depth and ego-motion estimation, leading to state-of-the-art results on benchmark datasets. However, one significant limitation shared by current methods is the assumption of a known parametric camera model -- usually the standard pinhole geometry -- leading to failure when applied to imaging systems that deviate significantly from this assumption (e.g., catadioptric cameras or underwater imaging). In this work, we show that self-supervision can be used to learn accurate depth and ego-motion estimation without prior knowledge of the camera model. Inspired by the geometric model of Grossberg and Nayar, we introduce Neural Ray Surfaces (NRS), convolutional networks that represent pixel-wise projection rays, approximating a wide range of cameras. NRS are fully differentiable and can be learned end-to-end from unlabeled raw videos. We demonstrate the use of NRS for self-supervised learning of visual odometry and depth estimation from raw videos obtained using a wide variety of camera systems, including pinhole, fisheye, and catadioptric.

Igor Vasiljevic, Nick Kolkin, Shanyi Zhang et al.

We introduce DIODE, a dataset that contains thousands of diverse high resolution color images with accurate, dense, long-range depth measurements. DIODE (Dense Indoor/Outdoor DEpth) is the first public dataset to include RGBD images of indoor and outdoor scenes obtained with one sensor suite. This is in contrast to existing datasets that focus on just one domain/scene type and employ different sensors, making generalization across domains difficult. The dataset is available for download at http://diode-dataset.org

Igor Vasiljevic, Ayan Chakrabarti, Gregory Shakhnarovich

State-of-the-art algorithms for many semantic visual tasks are based on the use of convolutional neural networks. These networks are commonly trained, and evaluated, on large annotated datasets of artifact-free high-quality images. In this paper, we investigate the effect of one such artifact that is quite common in natural capture settings: optical blur. We show that standard network models, trained only on high-quality images, suffer a significant degradation in performance when applied to those degraded by blur due to defocus, or subject or camera motion. We investigate the extent to which this degradation is due to the mismatch between training and input image statistics. Specifically, we find that fine-tuning a pre-trained model with blurred images added to the training set allows it to regain much of the lost accuracy. We also show that there is a fair amount of generalization between different degrees and types of blur, which implies that a single network model can be used robustly for recognition when the nature of the blur in the input is unknown. We find that this robustness arises as a result of these models learning to generate blur invariant representations in their hidden layers. Our findings provide useful insights towards developing vision systems that can perform reliably on real world images affected by blur.