We introduce a novel method for pre-training of large-scale vision encoders. Building on recent advancements in autoregressive pre-training of vision models, we extend this framework to a multimodal setting, i.e., images and text. In this paper, we present AIMV2, a family of generalist vision encoders characterized by a straightforward pre-training process, scalability, and remarkable performance across a range of downstream tasks. This is achieved by pairing the vision encoder with a multimodal decoder that autoregressively generates raw image patches and text tokens. Our encoders excel not only in multimodal evaluations but also in vision benchmarks such as localization, grounding, and classification. Notably, our AIMV2-3B encoder achieves 89.5% accuracy on ImageNet-1k with a frozen trunk. Furthermore, AIMV2 consistently outperforms state-of-the-art contrastive models (e.g., CLIP, SigLIP) in multimodal image understanding across diverse settings.

CLIP has been a celebrated method for training vision encoders to generate image/text representations facilitating various applications. Recently, it has been widely adopted as the vision backbone of multimodal large language models (MLLMs). The success of CLIP relies on aligning web-crawled noisy text annotations at image levels. However, such criteria may be insufficient for downstream tasks in need of fine-grained vision representations, especially when understanding region-level is demanding for MLLMs. We improve the localization capability of CLIP with several advances. Our proposed pre-training method, Contrastive Localized Language-Image Pre-training (CLOC), complements CLIP with region-text contrastive loss and modules. We formulate a new concept, promptable embeddings, of which the encoder produces image embeddings easy to transform into region representations given spatial hints. To support large-scale pre-training, we design a visually-enriched and spatially-localized captioning framework to effectively generate region-text labels. By scaling up to billions of annotated images, CLOC enables high-quality regional embeddings for recognition and retrieval tasks, and can be a drop-in replacement of CLIP to enhance MLLMs, especially on referring and grounding tasks.

Multi-label image classification is more applicable "in the wild" than single-label classification, as natural images usually contain multiple objects. However, exhaustively annotating images with every object of interest is costly and time-consuming. We train multi-label classifiers from datasets where each image is annotated with a single positive label only. As the presence of all other classes is unknown, we propose an Expected Negative loss that builds a set of expected negative labels in addition to the annotated positives. This set is determined based on prediction consistency, by averaging predictions over consecutive training epochs to build robust targets. Moreover, the ‘crop’ data augmentation leads to additional label noise by cropping out the single annotated object. Our novel spatial consistency loss improves supervision and ensures consistency of the spatial feature maps by maintaining per-class running-average heatmaps for each training image. We use MS-COCO, Pascal VOC, NUS-WIDE and CUB-Birds datasets to demonstrate the gains of the Expected Negative loss in combination with consistency and spatial consistency losses. We also demonstrate improved multi-label classification mAP on ImageNet-1K using the ReaL multi-label validation set.

We describe a method for real time video retrieval where the task is to match the 2D human pose of a query. A user can form a query by (i) interactively controlling a stickman on a web based GUI, (ii) uploading an image of the desired pose, or (iii) using the Kinect and acting out the query himself. The method is scalable and is applied to a dataset of 18 films totaling more than three million frames. The real time performance is achieved by searching for approximate nearest neighbors to the query using a random forest of K-D trees. Apart from the query modalities, we introduce two other areas of novelty. First, we show that pose retrieval can proceed using a low dimensional representation. Second, we show that the precision of the results can be improved substantially by combining the outputs of independent human pose estimation algorithms. The performance of the system is assessed quantitatively over a range of pose queries.

Most existing techniques for articulated Human Pose Estimation (HPE)consider each person independently. Here we tackle the problem in a new setting,coined Human Pose Coestimation (PCE), where multiple people are in a common,but unknown pose. The task of PCE is to estimate their poses jointly and toproduce prototypes characterizing the shared pose. Since the poses of the individual people should be similar to the prototype, PCE has less freedom compared to estimating each pose independently, which simplifies the problem.We demonstrate our PCE technique on two applications. The first is estimating the pose of people performing the same activity synchronously, such as during aerobics, cheerleading, and dancing in a group. We show that PCE improves pose estimation accuracy over estimating each person independently. The second application is learning prototype poses characterizing a pose class directly from an image search engine queried by the class name (e.g., “lotus pose”). We show that PCE leads to better pose estimation in such images, and it learns meaningful prototypes which can be used as priors for pose estimation in novel images.

We present a novel approach for estimating body part appearance models for pictorial structures. We learn latent relationships between the appearance of different body parts from annotated images, which then help in estimating better appearance models on novel images. The learned appearance models are general, in that they can be plugged into any pictorial structure engine. In a comprehensive evaluation we demonstrate the bene?ts brought by the new appearance models to an existing articulated human pose estimation algorithm, on hundreds of highly challenging images from the TV series Buffy the vampire slayer and the PASCAL VOC 2008 challenge.