About PRAW

Computer vision is rapidly emerging as the foundational intelligence layer for physical retail. The sector’s projected growth, from USD 4.23 billion in 2025 to USD 12.19 billion by 2030 (23.5% CAGR), underscores an accelerating demand for visual understanding in stores, warehouses, and last-mile environments.

Yet despite this momentum, physical retail remains one of the most demanding real-world testbeds for computer vision: unconstrained lighting, heavy occlusion, extreme intra-class similarity (e.g., thousands of near-identical SKUs), long-tailed distributions, and the need for real-time inference at the edge all push the boundaries of current methods.

Crucially, these systems must operate while maintaining a seamless customer experience: perception must be ambient, non-intrusive, and privacy-respecting, placing additional constraints on sensing modalities, model latency, and system design. Adding to this complexity, a new generation of platforms is bringing vision directly into physical retail - including inventory robots, smart carts, augmented reality devices, smart checkout stations - creating a new frontier of challenges.

This workshop will bridge academia and industry by convening researchers and practitioners who are advancing these topics in the retail domain, featuring invited talks from leading groups, peer-reviewed paper presentations, and a public challenge designed to promote reproducible evaluation on retail-specific benchmarks.

Call for Papers

We invite original research contributions addressing computer vision and artificial intelligence for physical retail and related real-world environments.

Topics of interest include, but are not limited to:

• Human behavior understanding in retail and physical environments, including for example action recognition, shopping activity understanding or people tracking
• Product detection, tracking, recognition, and retrieval for retail environments
• Scene understanding, indoor localization, and spatial reasoning in retail environments
• Vision-language, foundation, and generative AI models for retail applications
• Open-vocabulary, zero-/few-shot, continual, and open-world learning
• Efficient AI systems for real-world deployment, including edge inference, resource-constrained setups, and scalable installations
• Synthetic data generation, simulation, digital twins, datasets, benchmarks, and evaluation protocols

Submission Guidelines

We welcome submissions of regular papers following the same formatting and submission guidelines as the main conference. All accepted papers will be published in the official conference workshop proceedings.

Authors of accepted papers will be invited to present their work either as an oral presentation or as a poster during the workshop.

All submissions will undergo a double-blind peer-review process by at least three reviewers.

Challenge

The goal of this competition is to develop AI models that can accurately detect and localize shopper-shelf interactions in top-view retail videos.

Important Dates (GMT+2)

Invited Speakers (Tentative)

Alessia Saggese

University of Salerno

TDB

Dave Wollard

Standard AI

TBD

Vinod Kulathumani

Amazon

TBD

Arnab Dhua

Amazon

TBD

Tian Lan

Amazon

TBD

Program

PRAW 2027 will be a half-day workshop. The preliminary program includes:

The detailed schedule and presentation times will be announced later.

Program Committee

Organizers

Rocco Pietrini

Universitas Mercatorum

Marina Paolanti

University of Macerata

Davide Mazzini

Standard AI

Bruno Abbate

Standard AI

Dave Wollard

Standard AI

Vinod Kulathumani

Amazon

Yirui Wang

Amazon

Contact

For enquiries about submissions, the programme, the challenge, or collaboration, please contact:

rocco dot pietrini @ unimercatorum dot it