Amazon is proud to be a sponsor for the CVPR 2024 Social Event “Diversity and Inclusion for Everyone”, hosted by the organisers of Women in Computer Vision (WiCV) and LatinX in Computer Vision workshops.

Rapid urbanization poses social and environmental challenges. Addressing these issues effectively requires access to accurate and up-to-date 3D building models, obtained promptly and cost-effectively. Urban modeling is an interdisciplinary topic among computer vision, graphics, and photogrammetry. The demand for automated interpretation of scene geometry and semantics has surged due to various applications, including autonomous navigation, augmented reality, smart cities, and digital twins. As a result, substantial research effort has been dedicated to urban scene modeling within the computer vision and graphics communities, with a particular focus on photogrammetry, which has coped with urban modeling challenges for decades. This workshop is intended to bring researchers from these communities together. Through invited talks, spotlight presentations, a workshop challenge, and a poster session, it will increase interdisciplinary interaction and collaboration among photogrammetry, computer vision and graphics. We also solicit original contributions in the areas related to urban scene modeling.

Website: https://usm3d.github.io/

Featured Amazon keynote speakers: Ming Lin, Amazon Scholar; Sunil Hadap, Principal Applied Scientist

Website: https://vto-cvpr24.github.io/

The landscape of artificial intelligence is being transformed by the advent of Generative Foundation Models (GenFMs), such as Large Language Models (LLMs) and diffusion models. GenFMs offer unprecedented opportunities to enrich human lives and transform industries. However, they also pose significant challenges, including the generation of factually incorrect or biased information, which might be potentially harmful or misleading. With the emergence of multimodal GenFMs, which leverage and generate content in an increasing number of modalities, these challenges are set to become even more complex. This emphasizes the urgent need for rigorous and effective evaluation methodologies.

The 1st Workshop on Evaluation for Generative Foundation Models at CVPR 2024 aims to build a forum to discuss ongoing efforts in industry and academia, share best practices, and engage the community in working towards more reliable and scalable approaches for GenFMs evaluation.

Website: https://evgenfm.github.io/

Website: https://sites.google.com/view/fgvc11/

Website: https://generative-vision.github.io/workshop-CVPR-24/

Website: https://grocery-vision.github.io/cvpr2024.html

Website: https://cvpr.thecvf.com/virtual/2024/workshop/23586

This workshop aims to provide a platform for pioneers in prompting for vision to share recent advancements, showcase novel techniques and applications, and discuss open research questions about how the strategic use of prompts can unlock new levels of adaptability and performance in computer vision.

Website: https://prompting-in-vision.github.io/index_cvpr24.html

Website: https://opensun3d.github.io/index_cvpr24.html

Website: https://mula-workshop.github.io

The rapid development in computer vision and machine learning has caused a major disruption in the retail industry in recent years. In addition to the rise of online shopping, traditional markets also quickly embraced AI-related technology solutions at the physical store level. Following the introduction of computer vision to the world of retail, a new set of challenges emerged. These challenges were further expanded with the introduction of image and video generation capabilities.

The physical domain exhibits challenges such as the detection of shopper and product interactions, fine-grained recognition of visually similar products, as well as new products that are introduced on a daily basis. The online domain contains similar challenges, but with their own twist. Product search and recognition is performed on more than 100,000 classes, each including images, textual captions, and text by users during their search. In addition to discriminative machine learning, image generation has also started being used for the generation of product images and virtual try-on.

All of these challenges are shared by different companies in the field, and are also at the heart of the computer vision community. This workshop aims to present the progress in these challenges and encourage the forming of a community for retail computer vision.

Website: https://retailvisionworkshop.github.io/

Responsible Generative AI (ReGenAI) workshop aims to bring together researchers, practitioners, and industry leaders working at the intersection of generative AI, data, ethics, privacy and regulation, with the goal of discussing existing concerns, and brainstorming possible avenues forward to ensure the responsible progress of generative AI. We hope that the topics addressed in this workshop will constitute a crucial step towards ensuring a positive experience with generative AI for everyone.

Website: https://sites.google.com/view/cvpr-responsible-genai/home

Visual odometry and localization maintain an increasing interest in recent years, especially with the extensive applications on autonomous driving, augmented reality, and mobile computing. With the location information obtained through odometry, services based on location clues are also rapidly emerging. Particularly, in this workshop, we focus on mobile platform applications.

Website: https://sites.google.com/view/vocvalc2024

Wesbite: https://sites.google.com/view/2nd-mmfm-workshop/home?authuser=0

June 20-21, 11-11:30am

Amazon Lens is a feature which allows customers to search for products using their photos or live camera.

View in Your Room allows customers to preview how products like furniture would look in their home using Augmented reality.
Both features are available in the Amazon Mobile Shopping App today for anyone to use. We have videos showcasing these features available to show on conference displays, and team members can guide conference attendees to try the features out on their own devices.

June 19: 11:30am-12:00pm, 2:30-3pm
June 20: 11:30am-12:00pm, 1-1:30pm, 2:30-3pm
June 21: 11:30am-12:00pm, 1-1:30pm

Learn how Amazon Dash Cart and Amazon One are helping customers saving money, time and effort shopping for everyday grocery at scale, through computer vision and artificial intelligence! The Dash Cart is a smart cart that makes grocery trips faster and more personalized than ever. Find items quickly and easily. Add, remove, and weigh items right in your Dash Cart. When you're done shopping, skip the checkout line and roll out to your car. For more information, visit: https://aws.amazon.com/dash-cart/

June 19 12-12:30pm
June 21 12:30-1pm

Proteus is Amazon's first fully autonomous mobile robot. Historically, it’s been difficult to safely incorporate robotics where people are working in the same physical space as the robot. We believe Proteus will change that while remaining smart, safe, and collaborative.

June 19-20, 12:30-1:00pm

Making the most of your data often means using multiple AWS services. In this demo, learn about the zero-ETL integrations available for AWS Databases with AWS Analytics services and how they remove the need for you to build and manage complex data pipelines. Deep dive with a demo on how you can build your own pipeline with Amazon DynamoDB zero-ETL integration with Amazon OpenSearch.

June 19, 11-11:30am
June 20, 1:30-2:00pm

Retrieval Augmented Generation (RAG) helps improve the accuracy of outputs from Large Language Models (LLMs) by retrieving information from authoritative, predetermined knowledge sources. However, baseline RAG may flounder when a query requires connecting disparate information or a higher-level understanding of large data sets. GraphRAG combines the power of knowledge graphs and RAG technology to improve your generative AI application’s ability to answer questions across data sets, summarize concepts across a broad corpus, and provide human readable explanations of the results, therefore, improve its accuracy and reducing hallucinations. In this flash talk, learn how to use Amazon Neptune, our high-performance graph analytics and serverless database, to get started with GraphRAG and improve the accuracy of your generative AI applications.

June 19-20, 2-2:30pm

Generative AI and Foundational Models (FMs) are powerful technologies for building richer, personalized applications. With pgvector on Amazon Aurora PostgreSQL-Compatible Edition, you can access vector database capabilities to store, search, index, and query ML embeddings. Aurora is available as a Knowledge Base for Amazon Bedrock to securely connect your organization’s private data sources to FMs and enable Retrieval Augmented Generation (RAG) workflows on them. With Amazon Aurora Optimized Reads, you can boost vector search performance by up to 9x for memory-intensive workloads. In this demo, learn to integrate Aurora with Bedrock and how to utilize Optimized Reads to improve generative AI application performance.

June 19-20, 3-3:30pm

Serverless databases free you from capacity management while providing you with the economics of pay-per-use pricing. With AWS, customers have a broad choice of serverless databases to choose from, such as Amazon Aurora, Amazon DynamoDB, Amazon Neptune, and most recently Amazon ElastiCache. In this demo, learn how you can begin to instantly scale your own databases with Amazon ElastiCache Serverless and how to utilize the feature with the new open source project Valkey.

June 19, 3:30-4:00pm

Feedback from employees led us to create Amazon Robotics Identification (AR ID), an AI-powered scanning capability with innovative computer vision and machine learning technology to enable easier scanning of packages in our facilities. Currently, all packages in our facilities are scanned at each destination on their journey. In fulfillment centers, this scanning is currently manual—an item arrives at a workstation, the package is picked from a bin by an employee, and using a hand scanner, the employee finds the bar code and hand-scans the item.

AR ID removes the manual scanning process by using a unique camera system that runs at 120 frames per second, giving employees greater mobility and helping reduce the risk of injury. Employees can handle the packages freely with both hands instead of one hand while holding a scanner in the other, or they can work to position the package to scan it by hand. This creates a natural movement, and the technology does its job in the background.