Gomme Rinat Oleynick

About

Object Detection

The object detection AI ability (in Azure Video Indexer) identifies and tracks objects in video content, providing details like timestamps, confidence scores, and thumbnails. It helps users analyze and locate key moments, making video content more searchable and actionable.

Explore the Azure Video Indexer

Company

Microsoft

Type

Product Design

Service

UX UI Design

Year

2023

Motivation

Alongside the extensive data that Video Indexer can extract and identify (e.g., faces, brands, emotions, and more), we have received considerable feedback from users regarding the lack of an object detection feature.

The Goal

The challenge was to incorporate this new capability, optimize it functionally, and present it in a way that is clear to the user while seamlessly integrating with the existing visual language of the product without disrupting it.

Impact

Object detection feature is advancing video analytics and is projected to see significant adoption. AI powered tools like this, designed with users in mind, are becoming essential for efficiency and innovation across industries.

My Role

"I led the design and user experience efforts for the object detection feature, conducting extensive user research and usability testing at every stage of development. This process involved close collaboration with a multidisciplinary product team, including co-designers, UX researchers, data scientists, product managers, developers, and other stakeholders."

Results

Visual Detection

In addition to summarizing the data in the side panel, it was important to enable the highlighting of objects within the video itself and prominently reflect the selected object from the side panel.

Insights and Data

The main focus of Video Indexer is the data extracted from the video rather than the video itself. We’ve added a panel that displays and summarizes both the types of objects and their variety, while reflecting the time each object is highlighted.

Confidence

When utilizing AI-based capabilities, reliability is crucial. Therefore, it's important to provide users with a reflection of the system's confidence level in identifying an object as such, in addition to its representation on the timeline (a chronological display of data extracted from the video).

Focusing on Interests

Users may not necessarily want all objects present in the video to appear visually, as they likely have more specific intentions and needs. To avoid visual clutter, it was important for us to enable filtering and customization to suit their preferences.

Credits

Co-designers: Avi Neeman, Tal Nistor.

UX researcher: Sharon Yogev Maday.

UX writer: Merav Guttman.

Credits

This project was one of several I worked on at Microsoft as part of my roles in the Azure Video Indexer and AppSource Marketplace product teams. If you have any questions or would like to hear about other projects, feel free to reach out!

2024.

gomme.rinat@gmail.com