arXiv.itsnick.co

an independent field guide to 23 million papers from arXiv & OpenAlex — not affiliated with arXiv.org

Author

Sean Bell

31 papers · 2002–2026

2002: 12003: 42004: 22005: 02006: 02007: 32008: 12009: 02010: 02011: 02012: 02013: 12014: 22015: 42016: 12017: 12018: 02019: 02020: 22021: 02022: 22023: 32024: 22025: 12026: 1
2002peak 4 in 20032026

Frequent co-authors

Papers

showing 1–25 of 31 RSS

The Llama 3 Herd of Models

Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri +558 more

cs.AI · · cited 19,265× · arXiv ↗ · more like this ·

Modern artificial intelligence (AI) systems are powered by foundation models. This paper presents a new set of foundation models, called Llama 3. It is a herd of language models that natively support multilinguality, coding, reasoning, and tool usage. Our largest model is a dense Transformer with 405B parameters and a context window of up to 128K tokens. This paper presents an extensive empirical evaluation of Llama 3. We find that Llama 3 delivers comparable quality to leading language models such as GPT-4 on a plethora of tasks. We publicly release Llama 3, including pre-trained and post-trained versions of the 405B parameter language model and our Llama Guard 3 model for input and output safety. The paper also presents the results of experiments in which we integrate image, video, and speech capabilities into Llama 3 via a compositional approach. We observe this approach performs…

Inside-Outside Net: Detecting Objects in Context with Skip Pooling and Recurrent Neural Networks

Sean Bell, C. Lawrence Zitnick, Kavita Bala +1 more

cs.CV · · cited 1,275× · arXiv ↗ · more like this ·

It is well known that contextual and multi-scale representations are important for accurate visual recognition. In this paper we present the Inside-Outside Net (ION), an object detector that exploits information both inside and outside the region of interest. Contextual information outside the region of interest is integrated using spatial recurrent neural networks. Inside, we use skip pooling to extract information at multiple scales and levels of abstraction. Through extensive experiments we evaluate the design space and provide readers with an overview of what tricks of the trade are important. ION improves state-of-the-art on PASCAL VOC 2012 object detection from 73.9% to 76.4% mAP. On the new and more challenging MS COCO dataset, we improve state-of-art-the from 19.7% to 33.1% mAP. In the 2015 MS COCO Detection Challenge, our ION model won the Best Student Entry and finished 3rd…

Material Recognition in the Wild with the Materials in Context Database

Sean Bell, Paul Upchurch, Noah Snavely +1 more

cs.CV · · cited 593× · arXiv ↗ · more like this ·

Recognizing materials in real-world images is a challenging task. Real-world materials have rich surface texture, geometry, lighting conditions, and clutter, which combine to make the problem particularly difficult. In this paper, we introduce a new, large-scale, open dataset of materials in the wild, the Materials in Context Database (MINC), and combine this dataset with deep learning to achieve material recognition and segmentation of images in the wild. MINC is an order of magnitude larger than previous material databases, while being more diverse and well-sampled across its 23 categories. Using MINC, we train convolutional neural networks (CNNs) for two tasks: classifying materials from patches, and simultaneous material recognition and segmentation in full images. For patch-based classification on MINC we found that the best performing CNN architectures can achieve 85.2% mean class…

Learning visual similarity for product design with convolutional neural networks

Sean Bell, Kavita Bala

Computer Vision and Pattern Recognition · ACM Transactions on Graphics · · cited 426× · DOI ↗ · more like this ·

Popular sites like Houzz, Pinterest, and LikeThatDecor, have communities of users helping each other answer questions about products in images. In this paper we learn an embedding for visual search in interior design. Our embedding contains two different domains of product images: products cropped from internet scenes, and products in their iconic form. With such a multi-domain embedding, we demonstrate several applications of visual search including identifying products in scenes and finding stylistically similar products. To obtain the embedding, we train a convolutional neural network on pairs of images. We explore several training architectures including re-purposing object classifiers, using siamese networks, and using multitask learning. We evaluate our search quantitatively and qualitatively and demonstrate high quality results for search across multiple visual domains, enabling…

Intrinsic images in the wild

Sean Bell, Kavita Bala, Noah Snavely

Computer Vision and Pattern Recognition · ACM Transactions on Graphics · · cited 403× · DOI ↗ · more like this ·

Intrinsic image decomposition separates an image into a reflectance layer and a shading layer. Automatic intrinsic image decomposition remains a significant challenge, particularly for real-world scenes. Advances on this longstanding problem have been spurred by public datasets of ground truth data, such as the MIT Intrinsic Images dataset. However, the difficulty of acquiring ground truth data has meant that such datasets cover a small range of materials and objects. In contrast, real-world scenes contain a rich range of shapes and materials, lit by complex illumination. In this paper we introduce Intrinsic Images in the Wild , a large-scale, public dataset for evaluating intrinsic image decompositions of indoor scenes. We create this benchmark through millions of crowdsourced annotations of relative comparisons of material properties at pairs of points in each scene. Crowdsourcing…

Learning Visual Clothing Style with Heterogeneous Dyadic Co-occurrences

Andreas Veit, Balazs Kovacs, Sean Bell +3 more

cs.CV · · cited 334× · arXiv ↗ · more like this ·

With the rapid proliferation of smart mobile devices, users now take millions of photos every day. These include large numbers of clothing and accessory images. We would like to answer questions like `What outfit goes well with this pair of shoes?' To answer these types of questions, one has to go beyond learning visual similarity and learn a visual notion of compatibility across categories. In this paper, we propose a novel learning framework to help answer these types of questions. The main idea of this framework is to learn a feature transformation from images of items into a latent space that expresses compatibility. For the feature transformation, we use a Siamese Convolutional Neural Network (CNN) architecture, where training examples are pairs of items that are either compatible or incompatible. We model compatibility based on co-occurrence in large-scale user behavior data; in…

OpenSurfaces

Sean Bell, Paul Upchurch, Noah Snavely +1 more

Computer Graphics and Computer-Aided Design · ACM Transactions on Graphics · · cited 217× · DOI ↗ · more like this ·

The appearance of surfaces in real-world scenes is determined by the materials, textures, and context in which the surfaces appear. However, the datasets we have for visualizing and modeling rich surface appearance in context, in applications such as home remodeling, are quite limited. To help address this need, we present OpenSurfaces, a rich, labeled database consisting of thousands of examples of surfaces segmented from consumer photographs of interiors, and annotated with material parameters (reflectance, material names), texture information (surface normals, rectified textures), and contextual information (scene category, and object names). Retrieving usable surface information from uncalibrated Internet photo collections is challenging. We use human annotations and present a new methodology for segmenting and annotating materials in Internet photo collections suitable for…

‘Break a Leg—It’s all in the mind’: Police Officers’ Attitudes towards Colleagues with Mental Health Issues

Sean Bell, Yarin Eski

Political Science and International Relations · Policing A Journal of Policy and Practice · · cited 82× · DOI ↗ · more like this ·

Much has been written about police officer attitudes and interactions with members of the public experiencing mental health issues. However, little has been written or researched regarding police officers attitudes to colleagues with mental health issues. Within policing there is a culture that makes it difficult to discuss psychological injury and mental health with colleagues and managers. To do so is often seen as career destroying. The inherent cynicism associated with policing, lack of empathy, and macho culture further impedes discussion and ultimately access to support services. Austerity has reduced police training. Inadequate training results in officers not understanding mental health issues. There is an argument that personnel policies and systems victimizes officers and fails to understand their needs.

Tell Me What Happened: Unifying Text-guided Video Completion via Multimodal Masked Video Generation

Tsu-Jui Fu, Licheng Yu, Ning Zhang +4 more

cs.CV · · cited 44× · arXiv ↗ · more like this ·

Generating a video given the first several static frames is challenging as it anticipates reasonable future frames with temporal coherence. Besides video prediction, the ability to rewind from the last frame or infilling between the head and tail is also crucial, but they have rarely been explored for video completion. Since there could be different outcomes from the hints of just a few frames, a system that can follow natural language to perform video completion may significantly improve controllability. Inspired by this, we introduce a novel task, text-guided video completion (TVC), which requests the model to generate a video from partial frames guided by an instruction. We then propose Multimodal Masked Video Generation (MMVG) to address this TVC task. During training, MMVG discretizes the video frames into visual tokens and masks most of them to perform video completion from any…

GrokNet

Sean Bell, Yiqun Liu, Sami Alsheikh +6 more

Computer Vision and Pattern Recognition · · cited 27× · DOI ↗ · more like this ·

In this paper, we present GrokNet, a deployed image recognition system for commerce applications. GrokNet leverages a multi-task learning approach to train a single computer vision trunk. We achieve a 2.1x improvement in exact product match accuracy when compared to the previous state-of-the-art Facebook product recognition system. We achieve this by training on 7 datasets across several commerce verticals, using 80 categorical loss functions and 3 embedding losses. We share our experience of combining diverse sources with wide-ranging label semantics and image statistics, including learning from human annotations, user-generated tags, and noisy search engine interaction data. GrokNet has demonstrated gains in production applications and operates at Facebook scale.

Using Institutional Resources and Agency to Support Graduate Students’ Success at a Hispanic Serving Institution

Natalie Tran, Gaëtane Jean‐Marie, Katherine Powers +2 more

Education · Education Sciences · · cited 15× · DOI ↗ · more like this ·

There is a growing body of evidence that links increased social capital to minority student success in college. This paper seeks to expand specifically on the graduate experience of underrepresented minorities (URM) at a Hispanic Serving Institution (HSI) using the social capital framework. In a cross-sectional survey, 198 graduate students retrospectively considered the role of institutional resources and agents in their success towards graduation. Data revealed that motivational factors such as a sense of personal achievement, family support, peer support, career promotion, supportive faculty, program satisfaction, and faculty mentor played critical roles in the success of graduate students at HSI. Specifically, Latino students are more likely to report that faculty mentors played a significant role in their success compared to their non-Latino peers χ2(1, N = 195) = 5.33, p = 0.02.…

Shading Annotations in the Wild

Balazs Kovacs, Sean Bell, Noah Snavely +1 more

cs.CV · · cited 5× · arXiv ↗ · more like this ·

Understanding shading effects in images is critical for a variety of vision and graphics problems, including intrinsic image decomposition, shadow removal, image relighting, and inverse rendering. As is the case with other vision tasks, machine learning is a promising approach to understanding shading - but there is little ground truth shading data available for real-world images. We introduce Shading Annotations in the Wild (SAW), a new large-scale, public dataset of shading annotations in indoor scenes, comprised of multiple forms of shading judgments obtained via crowdsourcing, along with shading annotations automatically generated from RGB-D imagery. We use this data to train a convolutional neural network to predict per-pixel shading information in an image. We demonstrate the value of our data and network in an application to intrinsic images, where we can reduce decomposition…

Learning and Verification of Task Structure in Instructional Videos

Medhini Narasimhan, Licheng Yu, Sean Bell +2 more

cs.CV · · cited 4× · arXiv ↗ · more like this ·

Given the enormous number of instructional videos available online, learning a diverse array of multi-step task models from videos is an appealing goal. We introduce a new pre-trained video model, VideoTaskformer, focused on representing the semantics and structure of instructional videos. We pre-train VideoTaskformer using a simple and effective objective: predicting weakly supervised textual labels for steps that are randomly masked out from an instructional video (masked step modeling). Compared to prior work which learns step representations locally, our approach involves learning them globally, leveraging video of the entire surrounding task as context. From these learned representations, we can verify if an unseen video correctly executes a given task, as well as forecast which steps are likely to be taken after a given step. We introduce two new benchmarks for detecting mistakes…

RoPAWS: Robust Semi-supervised Representation Learning from Uncurated Data

Sangwoo Mo, Jong-Chyi Su, Chih-Yao Ma +4 more

cs.LG · · cited 2× · arXiv ↗ · more like this ·

Semi-supervised learning aims to train a model using limited labels. State-of-the-art semi-supervised methods for image classification such as PAWS rely on self-supervised representations learned with large-scale unlabeled but curated data. However, PAWS is often less effective when using real-world unlabeled data that is uncurated, e.g., contains out-of-class data. We propose RoPAWS, a robust extension of PAWS that can work with real-world unlabeled data. We first reinterpret PAWS as a generative classifier that models densities using kernel density estimation. From this probabilistic perspective, we calibrate its prediction based on the densities of labeled and unlabeled data, which leads to a simple closed-form solution from the Bayes' rule. We demonstrate that RoPAWS significantly improves PAWS for uncurated Semi-iNat by +5.3% and curated ImageNet by +0.4%.

Maternity in policing: the supervisor lottery, structural barriers, and insights for organisational reform

Keely Duddin, Kendal E. Wright, Arun Sondhi +4 more

Political Science and International Relations · Policing & Society · · cited 0× · DOI ↗ · more like this ·

This study examines women's maternity experiences in UK policing across pregnancy, maternity leave, and return to work. Drawing on a large-scale mixed-methods survey (n = 4752), it highlights how organisational culture, rather than formal policy alone, shapes maternity journeys, wellbeing, and long-term career trajectories. Quantitative findings reveal declining support across the maternity stages, with only 25% of participants reporting adequate mental health support during pregnancy, and just 20% upon returning to work. Supervisor discretion, termed the ‘supervisor lottery’, emerged as a critical factor, with inconsistent communication, marginalisation of flexible working, and organisational silence during maternity leave undermining inclusion and engagement. Support experiences varied significantly across role, disability status, nationality, and length of service. Regression and…

Inpainting-Guided Policy Optimization for Diffusion Large Language Models

Siyan Zhao, Mengchen Liu, Jing Huang +8 more

cs.LG · · cited 0× · arXiv ↗ · more like this ·

Masked diffusion large language models (dLLMs) are emerging as promising alternatives to autoregressive LLMs, offering competitive performance while supporting unique generation capabilities such as inpainting. We explore how inpainting can inform RL algorithm design for dLLMs. Aligning LLMs with reinforcement learning faces an exploration challenge: sparse reward signals and sample waste when models fail to discover correct solutions. While this inefficiency affects LLMs broadly, dLLMs offer a distinctive opportunity--their inpainting ability can guide exploration. We introduce IGPO (Inpainting Guided Policy Optimization), an RL framework that strategically inserts partial ground-truth reasoning traces during online sampling. Unlike providing full solutions, inpainting steers exploration toward promising trajectory spaces while preserving self-generated reasoning, bridging supervised…

Policing and occupational cultures

Sean Bell, Nick Kealey

Political Science and International Relations · Policy Press eBooks · · cited 0× · DOI ↗ · more like this ·

This chapter examines police culture and so-called canteen culture in its different guises. On the one hand, culture is a negative feature of policing, potentially detrimental to the service and the communities they serve. On the other, it can be a safe place for officers and staff to make sense of the world they work in, where they can seek and offer support to colleagues suffering the stresses and strains of operational encounters, notwithstanding the loss of shared workspaces that can leave officers isolated. It goes on to examine the impact of poor mental health on officers and staff and how cultural barriers deter help seeking. The chapter also examines the impact of a changing workforce on police culture and how a younger, better-educated, more diverse recruitment pool can challenge the status quo.

Cursiter Quarry (Phase 2), Grimbister, Kirkwall, Orkney Archaeological Field Evaluation Data Structure Report

Sean Bell

Anthropology · Archaeology Data Service · · cited 0× · DOI ↗ · more like this ·

An archaeological field evaluation and standing structure recording was undertaken as part of the Phase 2 extension of Cursiter Quarry, Grimbister, Orkney The current archaeological works follow on from a walkover survey undertaken by ORCA in 2016 as part of a Historic Environment Desk Baseline Assessment, and a geophysical survey undertaken by Rose Geophysics in 2018. The survey identified a total of twenty-one anomalies within the proposed development area, fourteen of which were considered to be of 'possible archaeological origin'. Following consultation with the Local Authority Archaeologist, and in line with best practice, a planning condition was placed on the proposed development requiring the recording of a boundary wall identified by the Historic Environment Desk Baseline Assessment along with a programme of trial trenching to investigate the geophysical anomalies. The standing…

Greens, Toab, Deerness, Orkney, KW17 2QG: Archaeological Evaluation Data Structure Report.

Sean Bell

Anthropology · Archaeology Data Service · · cited 0× · DOI ↗ · more like this ·

An archaeological field evaluation was undertaken by ORCA Archaeology on behalf of Orkney Builders (Contractors) on a site of Greens, Toab, Deerness, Orkney prior to the construction of two domestic properties. In 1923 a Class I Pictish Symbol Stone (CANMORE ID 3064) was recovered close to the buildings on the Site. This flagstone bears incised decoration comprising a circular disc and rectangle with square indentation, crescent and V-rod with a mirror below. Following consultation with the Local Authority Archaeologist and in line with best practice, the field evaluation consisted of a programme of trial trenching with attendant building recording. The results of the fieldwork show that much of the Site had been disturbed during the Modern Period with a number of buildings being modified or demolished, with the rubble material used for levelling, and many of the deposits were heavily…

Archaeological Desk-Based Assessment, 5-7 Station Road, Conisbrough, South Yorkshire

Sean Bell

Anthropology · Archaeology Data Service · · cited 0× · DOI ↗ · more like this ·

ARCUS was commissioned by Rofos Design and Technical to undertake an archaeological desk-based assessment of 5-7 Station Road, Conisbrough, South Yorkshire (NGR SK 511 991). The assessment was required to support a planning application for a residential development. The desk-based assessment comprised a site visit, documentary and cartographic research.

An Archaeological Field Evaluation at Land between 7 and 15 Cotton Street, Bolsover, Derbyshire

Sean Bell

Anthropology · Archaeology Data Service · · cited 0× · DOI ↗ · more like this ·

Two trenches, each measuring 8.25m by 2.5m were excavated. An identical sequence of deposits were identified in both trenches, consisting of modern surfaces and concrete layers directly overlying the undisturbed limestone. Trench 1 was excavated to a maximum depth of168.47m AOD, and Trench 2 to a maximum depth of 168.61m AOD. The aims of the investigations were: *To determine the extent, condition, character, importance and date of any archaeological remains present; *To provide information that will enable the remains to be placed within their local, regional, and national context and for an assessment of the significance of the archaeology of the proposal area to be made; *To provide information to inform the Development Control Archaeologist as to the requirement for further archaeological fieldwork as part of any mitigation strategy. In both trenches a deposit of cream-coloured…

Archaeological Mitigation at Green Building (Phase 2c), Round Foundry, Holbeck, Leeds, West Yorkshire.

Sean Bell

Anthropology · Archaeology Data Service · · cited 0× · DOI ↗ · more like this ·

CTP St. James had been granted planning consent for the construction of a new building (the Green Building) on a site occupied by a car park with a former industrial site known as the Round Foundry in Holbeck, Leeds. Following an archaeological evaluation of the site, an archaeological condition was placed upon it by the West Yorkshire Archaeology Advisory Service (WYAAS). The condition required the archaeological mitigation of the deposits within the development area by excavation. ARCUS were commissioned by CTP ST James to undertake the archaeological mitigation at the site of the new Green Building (Phase 2c). This document comprises the Assessment Report of the mitigation fieldwork and gives a brief overview of the archaeological features and deposits exposed with a view to establishing their historical and archaeological significance. This document will be used by WYAAS to advise…

Archaeological Mitigation at Dry Sand Foundry, Round Foundry, Holbeck, Leeds, West Yorkshire.

Sean Bell, Susie Matthewson

Anthropology · Archaeology Data Service · · cited 0× · DOI ↗ · more like this ·

CTP St James had been granted planning consent for the shell repair of the former Dry Sand Foundry building. The planning consent had an archaeological condition placed upon it, at the request of West Yorkshire Archaeology Advisory Service, requiring the archaeological mitigation of the deposits within the footprint of the Dry Sand Foundry building as part of the work entailed the insertion of a mezzanine floor, resting on piles. This scheme had been devised by ARCUS to use specialist piling process which would cause minimal disturbance to the archaeological deposits and thus allow the majority of the archaeological deposits to remain in situ. This scheme was approved by West Yorkshire Archaeology Advisory Service and CTP St James. The piling scheme required the removal of material to a height of 27.42m above site datum (27.20mAOD) within the interior of the building which had been…

Archaeological trial trenching, machine strip and record exercise and watching brief at Northowram Hospital, Northowram, Halifax, West Yorkshire

Chris G Cumberpatch, Sean Bell, Richard O'Neill +2 more

Anthropology · Archaeology Data Service · · cited 0× · DOI ↗ · more like this ·

ARCUS were commissioned by Persimmon Homes (West Yorkshire) Limited to undertake a further programme of archaeological evaluation by trial trenching (Trenches 1-12), a machine strip and record exercise (Areas A and B), and a watching brief (Areas 1-3), on the site of a residential development at Northowram Hospital, Northowram, Halifax, West Yorkshire (centred on NGR SE 111 277). The archaeological works were required as a condition of planning approval. Northowram Hall is thought to have been constructed on the site of the deserted medieval settlement of Whithill, mentioned in the Wakefield Manor Court Rolls from the thirteenth century. Its exact location has not been defined, but it is thought to have occupied the southern half of the development, adjacent to the hall. A series of owners of the land were mentioned throughout the medieval and post-medieval periods. Previous evaluation…

An archaeological field evaluation at land adjoining North Baileygate Mill Dam Lane, Pontefract, West Yorkshire

Antonia Thomas, Chris G Cumberpatch, Sean Bell

Anthropology · Archaeology Data Service · · cited 0× · DOI ↗ · more like this ·

ARCUS was commissioned by Persimmon Homes (West Yorkshire) Limited to undertake an archaeological field evaluation on the site of a former animal by- products processing factory, North Baileygate, Pontefract, West Yorkshire. The archaeological works were required by WYAS Advisory Service in line with the Department of the Environments guidelines, PPG16, and followed previous archaeological works. These consisted of an archaeological desk-based assessment and the excavation of four field evaluation trenches. A total of five, machine-assisted trenches, numbered five to nine, were excavated. Material and deposits predating the modern period were identified in Trenches 5 and 9, close to the Mill Dam Lane frontage and along the North Baileygate frontage. Structures relating to a late nineteenth century industrial complex were identified within Trench 8, underlying the tallow refinery, which…