arXiv subjects

Lijun Yu

Publications and source records attributed to Lijun Yu.

30 records · Page 2Linked to original sources

arXiv2024

Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation

While Large Language Models (LLMs) are the dominant models for generative tasks in language, they do not perform as well as diffusion models on image and video generation. To effectively use LLMs for visual generation, one crucial component is the visual tokenizer that maps pixel-space inputs to discrete tokens appropriate for LLM learning. In this paper, we introduce MAGVIT-v2, a video tokenizer designed to generate concise and expressive tokens for both videos and images using a common token vocabulary. Equipped with this new tokenizer, we show that LLMs outperform diffusion models on standard image and video generation benchmarks including ImageNet and Kinetics. In addition, we demonstrate that our tokenizer surpasses the previously top-performing video tokenizer on two more tasks: (1) video compression comparable to the next-generation video codec (VCC) according to human evaluations, and (2) learning effective representations for action recognition tasks.

cs.CV↗

arXiv2023

Photorealistic Video Generation with Diffusion Models

We present W.A.L.T, a transformer-based approach for photorealistic video generation via diffusion modeling. Our approach has two key design decisions. First, we use a causal encoder to jointly compress images and videos within a unified latent space, enabling training and generation across modalities. Second, for memory and training efficiency, we use a window attention architecture tailored for joint spatial and spatiotemporal generative modeling. Taken together these design decisions enable us to achieve state-of-the-art performance on established video (UCF-101 and Kinetics-600) and image (ImageNet) generation benchmarks without using classifier free guidance. Finally, we also train a cascade of three models for the task of text-to-video generation consisting of a base latent video diffusion model, and two video super-resolution diffusion models to generate videos of $512 \times 896$ resolution at $8$ frames per second.

cs.CV↗

arXiv2023

SPAE: Semantic Pyramid AutoEncoder for Multimodal Generation with Frozen LLMs

In this work, we introduce Semantic Pyramid AutoEncoder (SPAE) for enabling frozen LLMs to perform both understanding and generation tasks involving non-linguistic modalities such as images or videos. SPAE converts between raw pixels and interpretable lexical tokens (or words) extracted from the LLM's vocabulary. The resulting tokens capture both the semantic meaning and the fine-grained details needed for visual reconstruction, effectively translating the visual content into a language comprehensible to the LLM, and empowering it to perform a wide array of multimodal tasks. Our approach is validated through in-context learning experiments with frozen PaLM 2 and GPT 3.5 on a diverse set of image understanding and generation tasks. Our method marks the first successful attempt to enable a frozen LLM to generate image content while surpassing state-of-the-art performance in image understanding tasks, under the same setting, by over 25%.

cs.CV↗

arXiv2023

DocumentNet: Bridging the Data Gap in Document Pre-Training

Document understanding tasks, in particular, Visually-rich Document Entity Retrieval (VDER), have gained significant attention in recent years thanks to their broad applications in enterprise AI. However, publicly available data have been scarce for these tasks due to strict privacy constraints and high annotation costs. To make things worse, the non-overlapping entity spaces from different datasets hinder the knowledge transfer between document types. In this paper, we propose a method to collect massive-scale and weakly labeled data from the web to benefit the training of VDER models. The collected dataset, named DocumentNet, does not depend on specific document types or entity sets, making it universally applicable to all VDER tasks. The current DocumentNet consists of 30M documents spanning nearly 400 document types organized in a four-level ontology. Experiments on a set of broadly adopted VDER tasks show significant improvements when DocumentNet is incorporated into the pre-training for both classic and few-shot learning settings. With the recent emergence of large language models (LLMs), DocumentNet provides a large data source to extend their multi-modal capabilities for VDER.

cs.CL↗

arXiv2023

MAGVIT: Masked Generative Video Transformer

We introduce the MAsked Generative VIdeo Transformer, MAGVIT, to tackle various video synthesis tasks with a single model. We introduce a 3D tokenizer to quantize a video into spatial-temporal visual tokens and propose an embedding method for masked video token modeling to facilitate multi-task learning. We conduct extensive experiments to demonstrate the quality, efficiency, and flexibility of MAGVIT. Our experiments show that (i) MAGVIT performs favorably against state-of-the-art approaches and establishes the best-published FVD on three video generation benchmarks, including the challenging Kinetics-600. (ii) MAGVIT outperforms existing methods in inference time by two orders of magnitude against diffusion models and by 60x against autoregressive models. (iii) A single MAGVIT model supports ten diverse generation tasks and generalizes across videos from different visual domains. The source code and trained models will be released to the public at https://magvit.cs.cmu.edu.

cs.CV↗

arXiv2023

Score-based Continuous-time Discrete Diffusion Models

Score-based modeling through stochastic differential equations (SDEs) has provided a new perspective on diffusion models, and demonstrated superior performance on continuous data. However, the gradient of the log-likelihood function, i.e., the score function, is not properly defined for discrete spaces. This makes it non-trivial to adapt \textcolor{\cdiff}{the score-based modeling} to categorical data. In this paper, we extend diffusion models to discrete variables by introducing a stochastic jump process where the reverse process denoises via a continuous-time Markov chain. This formulation admits an analytical simulation during backward sampling. To learn the reverse process, we extend score matching to general categorical data and show that an unbiased estimator can be obtained via simple matching of the conditional marginal distributions. We demonstrate the effectiveness of the proposed method on a set of synthetic and real-world music and image benchmarks.

cs.LG↗

arXiv2022

Comparing value of travel time and value of travel time saving with heterogeneity in travelers

In research on the value of past time, the value of travel time and the value of saving travel time are two different concepts that have been vaguely distinguished over an extended period of time. This paper applies the theory of perspectives to discuss differences in the value of travel time and savings in travel time between different respondent groups and under different models. To this end, this paper designs an RP-SP questionnaire for urban travel behaviour, and collects data on the travel preferences of 409 Guangzhou residents. By introducing potential profiling to capture the effects of heterogeneity between classes, the respondent group is divided into three categories according to its individual socio-economic characteristics such as age, income, educational attainment, etc. After that, the utility function of travelers is established, and the MNL model, MIXL model, S-MNL model, G-MNL model, LCL model and MM-MNL model are selected for the total sample and three types of groups to estimate the time value. This paper emphasizes the definitions of VTT and VTTS, and presents a method for estimating VTT and VTTS using LPA to identify traveler heterogeneity.

physics.soc-ph↗

arXiv2022

Argus++: Robust Real-time Activity Detection for Unconstrained Video Streams with Overlapping Cube Proposals

Activity detection is one of the attractive computer vision tasks to exploit the video streams captured by widely installed cameras. Although achieving impressive performance, conventional activity detection algorithms are usually designed under certain constraints, such as using trimmed and/or object-centered video clips as inputs. Therefore, they failed to deal with the multi-scale multi-instance cases in real-world unconstrained video streams, which are untrimmed and have large field-of-views. Real-time requirements for streaming analysis also mark brute force expansion of them unfeasible. To overcome these issues, we propose Argus++, a robust real-time activity detection system for analyzing unconstrained video streams. The design of Argus++ introduces overlapping spatio-temporal cubes as an intermediate concept of activity proposals to ensure coverage and completeness of activity detection through over-sampling. The overall system is optimized for real-time processing on standalone consumer-level hardware. Extensive experiments on different surveillance and driving scenarios demonstrated its superior performance in a series of activity detection benchmarks, including CVPR ActivityNet ActEV 2021, NIST ActEV SDL UF/KF, TRECVID ActEV 2020/2021, and ICCV ROAD 2021.

cs.CV↗

arXiv2021

Generation and dynamics of soliton and soliton molecules from a VSe2/GO-based fiber laser

Recently, in addition to exploring the application of new saturable absorber devices in fiber lasers, soliton dynamics has also become a focus of current research. In this article, we report an ultrashort pulse fiber laser based on VSe2/GO nanocomposite and verify the formation process of soliton and soliton molecules by the numerical simulation. The prepared VSe2/GO-based device shows excellent saturable absorption characteristics with a modulation depth of 14.3% and a saturation absorption intensity of 0.93MW/cm2. The conventional soliton is obtained with pulse width of 573fs, which is currently the narrowest pulse width based on VSe2-related material, and has a signal-to-noise ratio of 60.4 dB. In addition, the soliton molecules are realized based on the VSe2/GO for the first time and have a pulse interval of ~2.2ps. We study the soliton dynamics through numerical simulation and reveal that before the formation of the soliton, it undergoes multiple nonlinear stages, such as soliton mode-locking, soliton splitting, and soliton oscillation. Furthermore, the results of numerical simulation are agreed well with the experimental data. These results indicate that the VSe2/GO might be another promising saturable absorber material for ultrafast photonics, and deepen the understanding of soliton dynamics in ultrafast fiber lasers.

physics.optics↗

arXiv2020

Training-free Monocular 3D Event Detection System for Traffic Surveillance

We focus on the problem of detecting traffic events in a surveillance scenario, including the detection of both vehicle actions and traffic collisions. Existing event detection systems are mostly learning-based and have achieved convincing performance when a large amount of training data is available. However, in real-world scenarios, collecting sufficient labeled training data is expensive and sometimes impossible (e.g. for traffic collision detection). Moreover, the conventional 2D representation of surveillance views is easily affected by occlusions and different camera views in nature. To deal with the aforementioned problems, in this paper, we propose a training-free monocular 3D event detection system for traffic surveillance. Our system firstly projects the vehicles into the 3D Euclidean space and estimates their kinematic states. Then we develop multiple simple yet effective ways to identify the events based on the kinematic patterns, which need no further training. Consequently, our system is robust to the occlusions and the viewpoint changes. Exclusive experiments report the superior result of our method on large-scale real-world surveillance datasets, which validates the effectiveness of our proposed system.

cs.CV↗

arXiv2018

Traffic Danger Recognition With Surveillance Cameras Without Training Data

We propose a traffic danger recognition model that works with arbitrary traffic surveillance cameras to identify and predict car crashes. There are too many cameras to monitor manually. Therefore, we developed a model to predict and identify car crashes from surveillance cameras based on a 3D reconstruction of the road plane and prediction of trajectories. For normal traffic, it supports real-time proactive safety checks of speeds and distances between vehicles to provide insights about possible high-risk areas. We achieve good prediction and recognition of car crashes without using any labeled training data of crashes. Experiments on the BrnoCompSpeed dataset show that our model can accurately monitor the road, with mean errors of 1.80% for distance measurement, 2.77 km/h for speed measurement, 0.24 m for car position prediction, and 2.53 km/h for speed prediction.

cs.CV↗

arXiv2018

MOBA-Slice: A Time Slice Based Evaluation Framework of Relative Advantage between Teams in MOBA Games

Multiplayer Online Battle Arena (MOBA) is currently one of the most popular genres of digital games around the world. The domain of knowledge contained in these complicated games is large. It is hard for humans and algorithms to evaluate the real-time game situation or predict the game result. In this paper, we introduce MOBA-Slice, a time slice based evaluation framework of relative advantage between teams in MOBA games. MOBA-Slice is a quantitative evaluation method based on learning, similar to the value network of AlphaGo. It establishes a foundation for further MOBA related research including AI development. In MOBA-Slice, with an analysis of the deciding factors of MOBA game results, we design a neural network model to fit our discounted evaluation function. Then we apply MOBA-Slice to Defense of the Ancients 2 (DotA2), a typical and popular MOBA game. Experiments on a large number of match replays show that our model works well on arbitrary matches. MOBA-Slice not only has an accuracy 3.7% higher than DotA Plus Assistant at result prediction, but also supports the prediction of the remaining time of the game, and then realizes the evaluation of relative advantage between teams.

cs.LG↗