Haoqi Fan

Chaitanya Ryali, Yuan-Ting Hu, Daniel Bolya et al. · meta-ai

Modern hierarchical vision transformers have added several vision-specific components in the pursuit of supervised classification performance. While these components lead to effective accuracies and attractive FLOP counts, the added complexity actually makes these transformers slower than their vanilla ViT counterparts. In this paper, we argue that this additional bulk is unnecessary. By pretraining with a strong visual pretext task (MAE), we can strip out all the bells-and-whistles from a state-of-the-art multi-stage vision transformer without losing accuracy. In the process, we create Hiera, an extremely simple hierarchical vision transformer that is more accurate than previous models while being significantly faster both at inference and during training. We evaluate Hiera on a variety of tasks for image and video recognition. Our code and models are available at https://github.com/facebookresearch/hiera.

Chen Wei, Karttikeya Mangalam, Po-Yao Huang et al. · meta-ai

There has been a longstanding belief that generation can facilitate a true understanding of visual data. In line with this, we revisit generatively pre-training visual representations in light of recent interest in denoising diffusion models. While directly pre-training with diffusion models does not produce strong representations, we condition diffusion models on masked input and formulate diffusion models as masked autoencoders (DiffMAE). Our approach is capable of (i) serving as a strong initialization for downstream recognition tasks, (ii) conducting high-quality image inpainting, and (iii) being effortlessly extended to video where it produces state-of-the-art classification accuracy. We further perform a comprehensive study on the pros and cons of design choices and build connections between diffusion models and masked autoencoders.

Mannat Singh, Quentin Duval, Kalyan Vasudev Alwala et al. · meta-ai

This paper revisits the standard pretrain-then-finetune paradigm used in computer vision for visual recognition tasks. Typically, state-of-the-art foundation models are pretrained using large scale (weakly) supervised datasets with billions of images. We introduce an additional pre-pretraining stage that is simple and uses the self-supervised MAE technique to initialize the model. While MAE has only been shown to scale with the size of models, we find that it scales with the size of the training dataset as well. Thus, our MAE-based pre-pretraining scales with both model and data size making it applicable for training foundation models. Pre-pretraining consistently improves both the model convergence and the downstream transfer performance across a range of model scales (millions to billions of parameters), and dataset sizes (millions to billions of images). We measure the effectiveness of pre-pretraining on 10 different visual recognition tasks spanning image classification, video recognition, object detection, low-shot classification and zero-shot recognition. Our largest model achieves new state-of-the-art results on iNaturalist-18 (91.7%), ImageNet-ReaL (91.1%), 1-shot ImageNet-1k (63.6%), and zero-shot transfer on Food-101 (96.2%). Our study reveals that model initialization plays a significant role, even for web-scale pretraining with billions of images, and our models are available publicly.

Po-Yao Huang, Vasu Sharma, Hu Xu et al. · meta-ai, mit

We present Masked Audio-Video Learners (MAViL) to train audio-visual representations. Our approach learns with three complementary forms of self-supervision: (1) reconstruction of masked audio and video input data, (2) intra- and inter-modal contrastive learning with masking, and (3) self-training by reconstructing joint audio-video contextualized features learned from the first two objectives. Pre-training with MAViL not only enables the model to perform well in audio-visual classification and retrieval tasks but also improves representations of each modality in isolation, without using information from the other modality for fine-tuning or inference. Empirically, MAViL sets a new state-of-the-art on AudioSet (53.1 mAP) and VGGSound (67.1% accuracy). For the first time, a self-supervised audio-visual model outperforms ones that use external supervision on these benchmarks.

Karttikeya Mangalam, Haoqi Fan, Yanghao Li et al.

We present Reversible Vision Transformers, a memory efficient architecture design for visual recognition. By decoupling the GPU memory requirement from the depth of the model, Reversible Vision Transformers enable scaling up architectures with efficient memory usage. We adapt two popular models, namely Vision Transformer and Multiscale Vision Transformers, to reversible variants and benchmark extensively across both model sizes and tasks of image classification, object detection and video classification. Reversible Vision Transformers achieve a reduced memory footprint of up to 15.5x at roughly identical model complexity, parameters and accuracy, demonstrating the promise of reversible vision transformers as an efficient backbone for hardware resource limited training regimes. Finally, we find that the additional computational burden of recomputing activations is more than overcome for deeper models, where throughput can increase up to 2.3x over their non-reversible counterparts. Full code and trained models are available at https://github.com/facebookresearch/slowfast. A simpler, easy to understand and modify version is also available at https://github.com/karttikeya/minREV

Fan Ma, Mike Zheng Shou, Linchao Zhu et al.

As an important area in computer vision, object tracking has formed two separate communities that respectively study Single Object Tracking (SOT) and Multiple Object Tracking (MOT). However, current methods in one tracking scenario are not easily adapted to the other due to the divergent training datasets and tracking objects of both tasks. Although UniTrack \cite{wang2021different} demonstrates that a shared appearance model with multiple heads can be used to tackle individual tracking tasks, it fails to exploit the large-scale tracking datasets for training and performs poorly on single object tracking. In this work, we present the Unified Transformer Tracker (UTT) to address tracking problems in different scenarios with one paradigm. A track transformer is developed in our UTT to track the target in both SOT and MOT. The correlation between the target and tracking frame features is exploited to localize the target. We demonstrate that both SOT and MOT tasks can be solved within this framework. The model can be simultaneously end-to-end trained by alternatively optimizing the SOT and MOT objectives on the datasets of individual tasks. Extensive experiments are conducted on several benchmarks with a unified model trained on SOT and MOT datasets. Code will be available at https://github.com/Flowerfan/Trackron.

Christoph Feichtenhofer, Haoqi Fan, Yanghao Li et al.

This paper studies a conceptually simple extension of Masked Autoencoders (MAE) to spatiotemporal representation learning from videos. We randomly mask out spacetime patches in videos and learn an autoencoder to reconstruct them in pixels. Interestingly, we show that our MAE method can learn strong representations with almost no inductive bias on spacetime (only except for patch and positional embeddings), and spacetime-agnostic random masking performs the best. We observe that the optimal masking ratio is as high as 90% (vs. 75% on images), supporting the hypothesis that this ratio is related to information redundancy of the data. A high masking ratio leads to a large speedup, e.g., > 4x in wall-clock time or even more. We report competitive results on several challenging video datasets using vanilla Vision Transformers. We observe that MAE can outperform supervised pre-training by large margins. We further report encouraging results of training on real-world, uncurated Instagram data. Our study suggests that the general framework of masked autoencoding (BERT, MAE, etc.) can be a unified methodology for representation learning with minimal domain knowledge.

Yanghao Li, Haoqi Fan, Ronghang Hu et al.

We present Fast Language-Image Pre-training (FLIP), a simple and more efficient method for training CLIP. Our method randomly masks out and removes a large portion of image patches during training. Masking allows us to learn from more image-text pairs given the same wall-clock time and contrast more samples per iteration with similar memory footprint. It leads to a favorable trade-off between accuracy and training time. In our experiments on 400 million image-text pairs, FLIP improves both accuracy and speed over the no-masking baseline. On a large diversity of downstream tasks, FLIP dominantly outperforms the CLIP counterparts trained on the same data. Facilitated by the speedup, we explore the scaling behavior of increasing the model size, data size, or training length, and report encouraging results and comparisons. We hope that our work will foster future research on scaling vision-language learning.

Xiao Wang, Haoqi Fan, Yuandong Tian et al.

Many recent self-supervised frameworks for visual representation learning are based on certain forms of Siamese networks. Such networks are conceptually symmetric with two parallel encoders, but often practically asymmetric as numerous mechanisms are devised to break the symmetry. In this work, we conduct a formal study on the importance of asymmetry by explicitly distinguishing the two encoders within the network -- one produces source encodings and the other targets. Our key insight is keeping a relatively lower variance in target than source generally benefits learning. This is empirically justified by our results from five case studies covering different variance-oriented designs, and is aligned with our preliminary theoretical analysis on the baseline. Moreover, we find the improvements from asymmetric designs generalize well to longer training schedules, multiple other frameworks and newer backbones. Finally, the combined effect of several asymmetric designs achieves a state-of-the-art accuracy on ImageNet linear probing and competitive results on downstream transfer. We hope our exploration will inspire more research in exploiting asymmetry for Siamese representation learning.

Haoran You, Yunyang Xiong, Xiaoliang Dai et al.

Vision Transformers (ViTs) have shown impressive performance but still require a high computation cost as compared to convolutional neural networks (CNNs), one reason is that ViTs' attention measures global similarities and thus has a quadratic complexity with the number of input tokens. Existing efficient ViTs adopt local attention (e.g., Swin) or linear attention (e.g., Performer), which sacrifice ViTs' capabilities of capturing either global or local context. In this work, we ask an important research question: Can ViTs learn both global and local context while being more efficient during inference? To this end, we propose a framework called Castling-ViT, which trains ViTs using both linear-angular attention and masked softmax-based quadratic attention, but then switches to having only linear angular attention during ViT inference. Our Castling-ViT leverages angular kernels to measure the similarities between queries and keys via spectral angles. And we further simplify it with two techniques: (1) a novel linear-angular attention mechanism: we decompose the angular kernels into linear terms and high-order residuals, and only keep the linear terms; and (2) we adopt two parameterized modules to approximate high-order residuals: a depthwise convolution and an auxiliary masked softmax attention to help learn both global and local information, where the masks for softmax attention are regularized to gradually become zeros and thus incur no overhead during ViT inference. Extensive experiments and ablation studies on three tasks consistently validate the effectiveness of the proposed Castling-ViT, e.g., achieving up to a 1.8% higher accuracy or 40% MACs reduction on ImageNet classification and 1.2 higher mAP on COCO detection under comparable FLOPs, as compared to ViTs with vanilla softmax-based attentions.

Yuqing Wang, Zhijie Lin, Ceyuan Yang et al.

Unified multimodal models (UMMs) aim to handle perception and generation in a single model. Yet existing UMMs still rely on a frozen, separately pretrained VAE for image generation, imposing a structural bottleneck. Naively removing it introduces a quality gap, as the model must learn both high-level structure and low-level details from raw pixels. In this paper, we propose Representation Forcing (RF), a technique that closes this gap by making representation prediction a native capability of the model. Concretely, RF forces the decoder to autoregressively predict visual representations as intermediate tokens before pixels; these tokens then stay in context to guide pixel diffusion within the same backbone. By turning representations from perception outputs into generation targets, RF eliminates the need for any external generative latent space. We find that RF benefits both understanding and generation. On image generation, our pixel-space model with RF matches state-of-the-art VAE-based unified models. On image understanding, pixel-space RF generally outperforms its VAE-based variant. Together, these results offer an effective step toward end-to-end, bottleneck-free UMMs.

Shanchuan Lin, Ceyuan Yang, Zhijie Lin et al.

We propose continuous adversarial flow models, a type of continuous-time flow model trained with an adversarial objective. Unlike flow matching, which uses a fixed mean-squared-error criterion, our approach introduces a learned discriminator to guide training. This change in objective induces a different generalized distribution, which empirically produces samples that are better aligned with the target data distribution. Our method is primarily proposed for post-training existing flow-matching models, although it can also train models from scratch. On the ImageNet 256px generation task, our post-training substantially improves the guidance-free FID of latent-space SiT from 8.26 to 3.63 and of pixel-space JiT from 7.17 to 3.57. It also improves guided generation, reducing FID from 2.06 to 1.53 for SiT and from 1.86 to 1.80 for JiT. We further evaluate our approach on text-to-image generation, where it achieves improved results on both the GenEval and DPG benchmarks.

Chaorui Deng, Deyao Zhu, Kunchang Li et al.

Unifying multimodal understanding and generation has shown impressive capabilities in cutting-edge proprietary systems. In this work, we introduce BAGEL, an open-source foundational model that natively supports multimodal understanding and generation. BAGEL is a unified, decoder-only model pretrained on trillions of tokens curated from large-scale interleaved text, image, video, and web data. When scaled with such diverse multimodal interleaved data, BAGEL exhibits emerging capabilities in complex multimodal reasoning. As a result, it significantly outperforms open-source unified models in both multimodal generation and understanding across standard benchmarks, while exhibiting advanced multimodal reasoning abilities such as free-form image manipulation, future frame prediction, 3D manipulation, and world navigation. In the hope of facilitating further opportunities for multimodal research, we share the key findings, pretraining details, data creation protocal, and release our code and checkpoints to the community. The project page is at https://bagel-ai.org/

Yiyang Zhou, Haoqin Tu, Zijun Wang et al.

We propose MIRA, a new benchmark designed to evaluate models in scenarios where generating intermediate visual images is essential for successful reasoning. Unlike traditional CoT methods that rely solely on text, tasks in MIRA require models to generate and utilize intermediate images - such as sketches, structural diagrams, or path drawings - to guide their reasoning process. This setup closely mirrors how humans solve complex problems through "drawing to think". To solve this, MIRA focuses on tasks that are intrinsically challenging and involve complex structures, spatial relationships, or reasoning steps that are difficult to express through language alone. To ensure that our evaluation data is of high-quality, we include 546 multimodal problems, annotated with intermediate visual images and final answers. We also propose a unified evaluation protocol for MIRA that spans three levels of evaluation input: direct input with image and question only, text-only CoT input with image and thinking prompts, and Visual-CoT input with both annotated image clues and textual thinking prompts. To probe the upper bound of model capacity on our benchmark, we also report pass@k and majority voting accuracies under different k settings. Experimental results show that existing multimodal large language models, including strongest private models as well as strong open-weight models, perform poorly when relying solely on textual prompts. However, when intermediate visual cues are provided, model performance improves consistently, yielding an average relative gain of 33.7% across all models and tasks. We also probe the upper bound by expanding the search space and designing textual prompts aligned with Visual-CoT, but both yield only limited improvements compared to our Visual-CoT setting. These results underscore the critical role of imagined visual information in enabling successful reasoning on MIRA.

Zilong Huang, Qinghao Ye, Bingyi Kang et al.

We introduce SuperClass, a super simple classification method for vision-language pre-training on image-text data. Unlike its contrastive counterpart CLIP who contrast with a text encoder, SuperClass directly utilizes tokenized raw text as supervised classification labels, without the need for additional text filtering or selection. Due to the absence of the text encoding as contrastive target, SuperClass does not require a text encoder and does not need to maintain a large batch size as CLIP does. SuperClass demonstrated superior performance on various downstream tasks, including classic computer vision benchmarks and vision language downstream tasks. We further explored the scaling behavior of SuperClass on model size, training length, or data size, and reported encouraging results and comparisons to CLIP. https://github.com/x-cls/superclass

Dong Guo, Faming Wu, Feida Zhu et al. · pku

We present Seed1.5-VL, a vision-language foundation model designed to advance general-purpose multimodal understanding and reasoning. Seed1.5-VL is composed with a 532M-parameter vision encoder and a Mixture-of-Experts (MoE) LLM of 20B active parameters. Despite its relatively compact architecture, it delivers strong performance across a wide spectrum of public VLM benchmarks and internal evaluation suites, achieving the state-of-the-art performance on 38 out of 60 public benchmarks. Moreover, in agent-centric tasks such as GUI control and gameplay, Seed1.5-VL outperforms leading multimodal systems, including OpenAI CUA and Claude 3.7. Beyond visual and video understanding, it also demonstrates strong reasoning abilities, making it particularly effective for multimodal reasoning challenges such as visual puzzles. We believe these capabilities will empower broader applications across diverse tasks. In this report, we mainly provide a comprehensive review of our experiences in building Seed1.5-VL across model design, data construction, and training at various stages, hoping that this report can inspire further research. Seed1.5-VL is now accessible at https://www.volcengine.com/ (Volcano Engine Model ID: doubao-1-5-thinking-vision-pro-250428)

Chenhui Gou, Zilong Chen, Zeyu Wang et al.

This paper studies Visual Question-Visual Answering (VQ-VA): generating an image, rather than text, in response to a visual question -- an ability that has recently emerged in proprietary systems such as NanoBanana and GPT-Image. To also bring this capability to open-source models, we introduce VQ-VA World, a data-centric framework built around an agentic pipeline for large-scale, targeted data construction. Leveraging web-scale deployment, this pipeline crawls a massive amount of ~1.8M high-quality, interleaved image-text samples for model training. For evaluation, we further release IntelligentBench, a human-curated benchmark that systematically assesses VQ-VA along the aspects of world knowledge, design knowledge, and reasoning. Training with VQ-VA World data yields strong empirical gains: it helps LightFusion attain 53.06 on IntelligentBench, substantially surpassing the best prior open-source baselines (i.e., 7.78 from vanilla LightFusion; 1.94 from UniWorld-V1), and significantly narrowing the gap toward leading proprietary systems (e.g., 81.67 from NanoBanana; 82.64 from GPT-Image). By releasing the full suite of model weights, datasets, and pipelines, we hope to stimulate future research on VQ-VA.

Chao-Yuan Wu, Yanghao Li, Karttikeya Mangalam et al.

While today's video recognition systems parse snapshots or short clips accurately, they cannot connect the dots and reason across a longer range of time yet. Most existing video architectures can only process <5 seconds of a video without hitting the computation or memory bottlenecks. In this paper, we propose a new strategy to overcome this challenge. Instead of trying to process more frames at once like most existing methods, we propose to process videos in an online fashion and cache "memory" at each iteration. Through the memory, the model can reference prior context for long-term modeling, with only a marginal cost. Based on this idea, we build MeMViT, a Memory-augmented Multiscale Vision Transformer, that has a temporal support 30x longer than existing models with only 4.5% more compute; traditional methods need >3,000% more compute to do the same. On a wide range of settings, the increased temporal support enabled by MeMViT brings large gains in recognition accuracy consistently. MeMViT obtains state-of-the-art results on the AVA, EPIC-Kitchens-100 action classification, and action anticipation datasets. Code and models are available at https://github.com/facebookresearch/memvit.

Yanghao Li, Chao-Yuan Wu, Haoqi Fan et al.

In this paper, we study Multiscale Vision Transformers (MViTv2) as a unified architecture for image and video classification, as well as object detection. We present an improved version of MViT that incorporates decomposed relative positional embeddings and residual pooling connections. We instantiate this architecture in five sizes and evaluate it for ImageNet classification, COCO detection and Kinetics video recognition where it outperforms prior work. We further compare MViTv2s' pooling attention to window attention mechanisms where it outperforms the latter in accuracy/compute. Without bells-and-whistles, MViTv2 has state-of-the-art performance in 3 domains: 88.8% accuracy on ImageNet classification, 58.7 boxAP on COCO object detection as well as 86.1% on Kinetics-400 video classification. Code and models are available at https://github.com/facebookresearch/mvit.

Haoqi Fan, Tullie Murrell, Heng Wang et al.

We introduce PyTorchVideo, an open-source deep-learning library that provides a rich set of modular, efficient, and reproducible components for a variety of video understanding tasks, including classification, detection, self-supervised learning, and low-level processing. The library covers a full stack of video understanding tools including multimodal data loading, transformations, and models that reproduce state-of-the-art performance. PyTorchVideo further supports hardware acceleration that enables real-time inference on mobile devices. The library is based on PyTorch and can be used by any training framework; for example, PyTorchLightning, PySlowFast, or Classy Vision. PyTorchVideo is available at https://pytorchvideo.org/

Christoph Feichtenhofer, Haoqi Fan, Bo Xiong et al.

We present a large-scale study on unsupervised spatiotemporal representation learning from videos. With a unified perspective on four recent image-based frameworks, we study a simple objective that can easily generalize all these methods to space-time. Our objective encourages temporally-persistent features in the same video, and in spite of its simplicity, it works surprisingly well across: (i) different unsupervised frameworks, (ii) pre-training datasets, (iii) downstream datasets, and (iv) backbone architectures. We draw a series of intriguing observations from this study, e.g., we discover that encouraging long-spanned persistency can be effective even if the timespan is 60 seconds. In addition to state-of-the-art results in multiple benchmarks, we report a few promising cases in which unsupervised pre-training can outperform its supervised counterpart. Code is made available at https://github.com/facebookresearch/SlowFast

Haoqi Fan, Bo Xiong, Karttikeya Mangalam et al.

We present Multiscale Vision Transformers (MViT) for video and image recognition, by connecting the seminal idea of multiscale feature hierarchies with transformer models. Multiscale Transformers have several channel-resolution scale stages. Starting from the input resolution and a small channel dimension, the stages hierarchically expand the channel capacity while reducing the spatial resolution. This creates a multiscale pyramid of features with early layers operating at high spatial resolution to model simple low-level visual information, and deeper layers at spatially coarse, but complex, high-dimensional features. We evaluate this fundamental architectural prior for modeling the dense nature of visual signals for a variety of video recognition tasks where it outperforms concurrent vision transformers that rely on large scale external pre-training and are 5-10x more costly in computation and parameters. We further remove the temporal dimension and apply our model for image classification where it outperforms prior work on vision transformers. Code is available at: https://github.com/facebookresearch/SlowFast

Christoph Feichtenhofer, Haoqi Fan, Jitendra Malik et al.

We present SlowFast networks for video recognition. Our model involves (i) a Slow pathway, operating at low frame rate, to capture spatial semantics, and (ii) a Fast pathway, operating at high frame rate, to capture motion at fine temporal resolution. The Fast pathway can be made very lightweight by reducing its channel capacity, yet can learn useful temporal information for video recognition. Our models achieve strong performance for both action classification and detection in video, and large improvements are pin-pointed as contributions by our SlowFast concept. We report state-of-the-art accuracy on major video recognition benchmarks, Kinetics, Charades and AVA. Code has been made available at: https://github.com/facebookresearch/SlowFast

Chaorui Deng, Deyao Zhu, Kunchang Li et al.

We introduce Causal Diffusion as the autoregressive (AR) counterpart of Diffusion models. It is a next-token(s) forecasting framework that is friendly to both discrete and continuous modalities and compatible with existing next-token prediction models like LLaMA and GPT. While recent works attempt to combine diffusion with AR models, we show that introducing sequential factorization to a diffusion model can substantially improve its performance and enables a smooth transition between AR and diffusion generation modes. Hence, we propose CausalFusion - a decoder-only transformer that dual-factorizes data across sequential tokens and diffusion noise levels, leading to state-of-the-art results on the ImageNet generation benchmark while also enjoying the AR advantage of generating an arbitrary number of tokens for in-context reasoning. We further demonstrate CausalFusion's multimodal capabilities through a joint image generation and captioning model, and showcase CausalFusion's ability for zero-shot in-context image manipulations. We hope that this work could provide the community with a fresh perspective on training multimodal models over discrete and continuous data.

Ceyuan Yang, Zhijie Lin, Yang Zhao et al.

We present Omni, a unified multimodal model natively trained on diverse modalities, including text, images, videos, 3D geometry, and hidden representations. We find that such training enables Context Unrolling, where the model explicitly reasons across multiple modal representations before producing predictions. This process enables the model to aggregate complementary information across heterogeneous modalities, facilitating a more faithful approximation of the shared multimodal knowledge manifold and improving downstream reasoning fidelity. As a result, Omni achieves strong performance on both multimodal generation and understanding benchmarks, while demonstrating advanced multimodal reasoning capabilities, including in-context generation of text, image, video, and 3D geometry.

Qinghao Ye, Xianhan Zeng, Fu Li et al.

Image captioning has long been a pivotal task in visual understanding, with recent advancements in vision-language models (VLMs) significantly enhancing the ability to generate detailed image captions. However, the evaluation of detailed image captioning remains underexplored due to outdated evaluation metrics and coarse annotations. In this paper, we introduce DeCapBench along with a novel metric, DCScore, specifically designed for detailed captioning tasks. DCScore evaluates hallucinations and fine-grained comprehensiveness by deconstructing responses into the smallest self-sufficient units, termed primitive information units, and assessing them individually. Our evaluation shows that DCScore aligns more closely with human judgment than other rule-based or model-based metrics. Concurrently, DeCapBench exhibits a high correlation with VLM arena results on descriptive tasks, surpassing existing benchmarks for vision-language models. Additionally, we present an automatic fine-grained feedback collection method, FeedQuill, for preference optimization based on our advanced metric, showing robust generalization capabilities across auto-generated preference data. Extensive experiments on multiple VLMs demonstrate that our method not only significantly reduces hallucinations but also enhances performance across various benchmarks, achieving superior detail captioning performance while surpassing GPT-4o.

Shanchuan Lin, Ceyuan Yang, Zhijie Lin et al.

We present adversarial flow models, a class of generative models that unifies adversarial models and flow models. Our method supports native one-step or multi-step generation and is trained using the adversarial objective. Unlike traditional GANs, where the generator learns an arbitrary transport plan between the noise and the data distributions, our generator learns a deterministic noise-to-data mapping, which is the same optimal transport as in flow-matching models. This significantly stabilizes adversarial training. Also, unlike consistency-based methods, our model directly learns one-step or few-step generation without needing to learn the intermediate timesteps of the probability flow for propagation. This saves model capacity, reduces training iterations, and avoids error accumulation. Under the same 1NFE setting on ImageNet-256px, our B/2 model approaches the performance of consistency-based XL/2 models, while our XL/2 model creates a new best FID of 2.38. We additionally show the possibility of end-to-end training of 56-layer and 112-layer models through depth repetition without any intermediate supervision, and achieve FIDs of 2.08 and 1.94 using a single forward pass, surpassing their 2NFE and 4NFE counterparts.

Zeyu Wang, Zilong Chen, Chenhui Gou et al.

Unified multimodal models have recently shown remarkable gains in both capability and versatility, yet most leading systems are still trained from scratch and require substantial computational resources. In this paper, we show that competitive performance can be obtained far more efficiently by strategically fusing publicly available models specialized for either generation or understanding. Our key design is to retain the original blocks while additionally interleaving multimodal self-attention blocks throughout the networks. This double fusion mechanism (1) effectively enables rich multi-modal fusion while largely preserving the original strengths of the base models, and (2) catalyzes synergistic fusion of high-level semantic representations from the understanding encoder with low-level spatial signals from the generation encoder. By training with only ~ 35B tokens, this approach achieves strong results across multiple benchmarks: 0.91 on GenEval for compositional text-to-image generation, 82.16 on DPG-Bench for complex text-to-image generation, 6.06 on GEditBench, and 3.77 on ImgEdit-Bench for image editing. By fully releasing the entire suite of code, model weights, and datasets, we hope to support future research on unified multimodal modeling.

Chen Wei, Haoqi Fan, Saining Xie et al.

We present Masked Feature Prediction (MaskFeat) for self-supervised pre-training of video models. Our approach first randomly masks out a portion of the input sequence and then predicts the feature of the masked regions. We study five different types of features and find Histograms of Oriented Gradients (HOG), a hand-crafted feature descriptor, works particularly well in terms of both performance and efficiency. We observe that the local contrast normalization in HOG is essential for good results, which is in line with earlier work using HOG for visual recognition. Our approach can learn abundant visual knowledge and drive large-scale Transformer-based models. Without using extra model weights or supervision, MaskFeat pre-trained on unlabeled videos achieves unprecedented results of 86.7% with MViT-L on Kinetics-400, 88.3% on Kinetics-600, 80.4% on Kinetics-700, 39.8 mAP on AVA, and 75.0% on SSv2. MaskFeat further generalizes to image input, which can be interpreted as a video with a single frame and obtains competitive results on ImageNet.

Xitong Yang, Haoqi Fan, Lorenzo Torresani et al.

The standard way of training video models entails sampling at each iteration a single clip from a video and optimizing the clip prediction with respect to the video-level label. We argue that a single clip may not have enough temporal coverage to exhibit the label to recognize, since video datasets are often weakly labeled with categorical information but without dense temporal annotations. Furthermore, optimizing the model over brief clips impedes its ability to learn long-term temporal dependencies. To overcome these limitations, we introduce a collaborative memory mechanism that encodes information across multiple sampled clips of a video at each training iteration. This enables the learning of long-range dependencies beyond a single clip. We explore different design choices for the collaborative memory to ease the optimization difficulties. Our proposed framework is end-to-end trainable and significantly improves the accuracy of video classification at a negligible computational overhead. Through extensive experiments, we demonstrate that our framework generalizes to different video architectures and tasks, outperforming the state of the art on both action recognition (e.g., Kinetics-400 & 700, Charades, Something-Something-V1) and action detection (e.g., AVA v2.1 & v2.2).

Bo Xiong, Haoqi Fan, Kristen Grauman et al.

We present a multiview pseudo-labeling approach to video learning, a novel framework that uses complementary views in the form of appearance and motion information for semi-supervised learning in video. The complementary views help obtain more reliable pseudo-labels on unlabeled video, to learn stronger video representations than from purely supervised data. Though our method capitalizes on multiple views, it nonetheless trains a model that is shared across appearance and motion input and thus, by design, incurs no additional computation overhead at inference time. On multiple video recognition datasets, our method substantially outperforms its supervised counterpart, and compares favorably to previous work on standard benchmarks in self-supervised video representation learning.

Song Liu, Haoqi Fan, Shengsheng Qian et al.

Video-Text Retrieval has been a hot research topic with the growth of multimedia data on the internet. Transformer for video-text learning has attracted increasing attention due to its promising performance. However, existing cross-modal transformer approaches typically suffer from two major limitations: 1) Exploitation of the transformer architecture where different layers have different feature characteristics is limited; 2) End-to-end training mechanism limits negative sample interactions in a mini-batch. In this paper, we propose a novel approach named Hierarchical Transformer (HiT) for video-text retrieval. HiT performs Hierarchical Cross-modal Contrastive Matching in both feature-level and semantic-level, achieving multi-view and comprehensive retrieval results. Moreover, inspired by MoCo, we propose Momentum Cross-modal Contrast for cross-modal learning to enable large-scale negative sample interactions on-the-fly, which contributes to the generation of more precise and discriminative representations. Experimental results on the three major Video-Text Retrieval benchmark datasets demonstrate the advantages of our method.

Yutong Bai, Haoqi Fan, Ishan Misra et al.

Leveraging temporal information has been regarded as essential for developing video understanding models. However, how to properly incorporate temporal information into the recent successful instance discrimination based contrastive self-supervised learning (CSL) framework remains unclear. As an intuitive solution, we find that directly applying temporal augmentations does not help, or even impair video CSL in general. This counter-intuitive observation motivates us to re-design existing video CSL frameworks, for better integration of temporal knowledge. To this end, we present Temporal-aware Contrastive self-supervised learningTaCo, as a general paradigm to enhance video CSL. Specifically, TaCo selects a set of temporal transformations not only as strong data augmentation but also to constitute extra self-supervision for video understanding. By jointly contrasting instances with enriched temporal transformations and learning these transformations as self-supervised signals, TaCo can significantly enhance unsupervised video representation learning. For instance, TaCo demonstrates consistent improvement in downstream classification tasks over a list of backbones and CSL approaches. Our best model achieves 85.1% (UCF-101) and 51.6% (HMDB-51) top-1 accuracy, which is a 3% and 2.4% relative improvement over the previous state-of-the-art.

Xinlei Chen, Haoqi Fan, Ross Girshick et al.

Contrastive unsupervised learning has recently shown encouraging progress, e.g., in Momentum Contrast (MoCo) and SimCLR. In this note, we verify the effectiveness of two of SimCLR's design improvements by implementing them in the MoCo framework. With simple modifications to MoCo---namely, using an MLP projection head and more data augmentation---we establish stronger baselines that outperform SimCLR and do not require large training batches. We hope this will make state-of-the-art unsupervised learning research more accessible. Code will be made public.

Kaiming He, Haoqi Fan, Yuxin Wu et al.

We present Momentum Contrast (MoCo) for unsupervised visual representation learning. From a perspective on contrastive learning as dictionary look-up, we build a dynamic dictionary with a queue and a moving-averaged encoder. This enables building a large and consistent dictionary on-the-fly that facilitates contrastive unsupervised learning. MoCo provides competitive results under the common linear protocol on ImageNet classification. More importantly, the representations learned by MoCo transfer well to downstream tasks. MoCo can outperform its supervised pre-training counterpart in 7 detection/segmentation tasks on PASCAL VOC, COCO, and other datasets, sometimes surpassing it by large margins. This suggests that the gap between unsupervised and supervised representation learning has been largely closed in many vision tasks.

Yunpeng Chen, Haoqi Fan, Bing Xu et al.

In natural images, information is conveyed at different frequencies where higher frequencies are usually encoded with fine details and lower frequencies are usually encoded with global structures. Similarly, the output feature maps of a convolution layer can also be seen as a mixture of information at different frequencies. In this work, we propose to factorize the mixed feature maps by their frequencies, and design a novel Octave Convolution (OctConv) operation to store and process feature maps that vary spatially "slower" at a lower spatial resolution reducing both memory and computation cost. Unlike existing multi-scale methods, OctConv is formulated as a single, generic, plug-and-play convolutional unit that can be used as a direct replacement of (vanilla) convolutions without any adjustments in the network architecture. It is also orthogonal and complementary to methods that suggest better topologies or reduce channel-wise redundancy like group or depth-wise convolutions. We experimentally show that by simply replacing convolutions with OctConv, we can consistently boost accuracy for both image and video recognition tasks, while reducing memory and computational cost. An OctConv-equipped ResNet-152 can achieve 82.9% top-1 classification accuracy on ImageNet with merely 22.2 GFLOPs.

Chao-Yuan Wu, Christoph Feichtenhofer, Haoqi Fan et al.

To understand the world, we humans constantly need to relate the present to the past, and put events in context. In this paper, we enable existing video models to do the same. We propose a long-term feature bank---supportive information extracted over the entire span of a video---to augment state-of-the-art video models that otherwise would only view short clips of 2-5 seconds. Our experiments demonstrate that augmenting 3D convolutional networks with a long-term feature bank yields state-of-the-art results on three challenging video datasets: AVA, EPIC-Kitchens, and Charades.

Donghyun Yoo, Haoqi Fan, Vishnu Naresh Boddeti et al.

Feature representations from pre-trained deep neural networks have been known to exhibit excellent generalization and utility across a variety of related tasks. Fine-tuning is by far the simplest and most widely used approach that seeks to exploit and adapt these feature representations to novel tasks with limited data. Despite the effectiveness of fine-tuning, itis often sub-optimal and requires very careful optimization to prevent severe over-fitting to small datasets. The problem of sub-optimality and over-fitting, is due in part to the large number of parameters used in a typical deep convolutional neural network. To address these problems, we propose a simple yet effective regularization method for fine-tuning pre-trained deep networks for the task of k-shot learning. To prevent overfitting, our key strategy is to cluster the model parameters while ensuring intra-cluster similarity and inter-cluster diversity of the parameters, effectively regularizing the dimensionality of the parameter search space. In particular, we identify groups of neurons within each layer of a deep network that shares similar activation patterns. When the network is to be fine-tuned for a classification task using only k examples, we propagate a single gradient to all of the neuron parameters that belong to the same group. The grouping of neurons is non-trivial as neuron activations depend on the distribution of the input data. To efficiently search for optimal groupings conditioned on the input data, we propose a reinforcement learning search strategy using recurrent networks to learn the optimal group assignments for each network layer. Experimental results show that our method can be easily applied to several popular convolutional neural networks and improve upon other state-of-the-art fine-tuning based k-shot learning strategies by more than10%

Minghuang Ma, Haoqi Fan, Kris M. Kitani

We bring together ideas from recent work on feature design for egocentric action recognition under one framework by exploring the use of deep convolutional neural networks (CNN). Recent work has shown that features such as hand appearance, object attributes, local hand motion and camera ego-motion are important for characterizing first-person actions. To integrate these ideas under one framework, we propose a twin stream network architecture, where one stream analyzes appearance information and the other stream analyzes motion information. Our appearance stream encodes prior knowledge of the egocentric paradigm by explicitly training the network to segment hands and localize objects. By visualizing certain neuron activation of our network, we show that our proposed architecture naturally learns features that capture object attributes and hand-object configurations. Our extensive experiments on benchmark egocentric action datasets show that our deep architecture enables recognition rates that significantly outperform state-of-the-art techniques -- an average $6.6\%$ increase in accuracy over all datasets. Furthermore, by learning to recognize objects, actions and activities jointly, the performance of individual recognition tasks also increase by $30\%$ (actions) and $14\%$ (objects). We also include the results of extensive ablative analysis to highlight the importance of network design decisions..