Cognitive Image Recognition

Overview

This technical note explains Lexbe’s Cognitive Image Recognition service which uses artificial intelligence (“AI”)-based image recognition to automatically extract pertinent keywords based on objects that are displayed in an image.

Background

Some cases can contain large amounts of photos or other image-based evidence. Images traditionally have been challenging to review in document cases as they are not easily retrievable or searchable absent human coding, which can be expensive and time-consuming. Lexbe’s Cognitive Image Recognition service addresses this issue by automatically scanning supported images for thousands of objects that can be identified using cutting-edge AI. The identified objects are coded into the metadata for the case as machine-readable text for easy and accurate search, filtering, and retrieval.

Some case types routinely can involve large amounts of images as evidence. These include:

    • Construction Defect

    • Criminal

    • Environmental

    • Housing

    • Real Estate

Automatic image recognition can also be helpful in any case as pictures are routinely included as attachments in Custodian email. While routine attachments to emails may not have high evidentiary value as a matter of course, they may be hard to review efficiently as every normally must be viewed by a human reviewer and coded. With Lexbe’s Cognitive Image Recognition service, all images can be text searched and mass reviewed to determine whether they have evidentiary value our not.

Features

Objects Identified

Lexbe Cognitive Image Recognition identifies thousands of objects such as vehicles, buildings, people, and other objects within images that may be important in litigation. This makes it easy to search, filter, and code large document intensive collections with image-based evidence. Below is a sample:

    • Nature and Outdoors: ‘Beach’, ‘Mountains’, ‘Lake’, ‘Sunset’, ‘Rainbow’, etc.

    • Home and Garden: ‘Bed’, ‘Table’, ‘Backyard’, ‘Chandelier’, ‘Bedroom’, etc.

    • Plants and Flowers: ‘Rose’, ‘Tulip’, ‘Palm Tree’, ‘Forest’, ‘Bamboo’, etc.

    • Transportation and Vehicles: ‘Airplane’, ‘Car’, ‘Bicycle’, ‘Motorcycle’, ‘Truck’, etc.

    • Animals and Pets: ‘Dog’, ‘Cat’, ‘Horse’, ‘Tiger’, ‘Turtle’, etc.

    • Sports and Leisure: ‘Golf’, ‘Basketball’, ‘Hockey’, ‘Tennis’, ‘Hiking’, etc.

    • People and Events: ‘Wedding’, ‘Bride’, ‘Baby’, ‘Birthday Cake’, ‘Guitarist’, etc.

    • Food and Drink: ‘Apple’, ‘Sandwich’, ‘Wine’, ‘Cake’, ‘Pizza’, etc.

    • Art and Entertainment: ‘Sculpture’, ‘Painting’, ‘Guitar’, ‘Ballet’, ‘Mosaic’, etc.

    • Electronics: ‘Computer’, ‘Mobile Phone’, ‘Video Camera’, ‘TV’, ‘Headphones’, etc.

Text ID

Lexbe Cognitive Image Recognition also extracts text within images, including text in natural scenes such as road signs or license plates, text over objects such as t-shirts or mugs, and text on screen such as captions or news. This AI-based image recognition is much more accurate than traditional OCR.

Supported File Types

The Cognitive Image Recognition service supports the following image file types:

    • BMP

    • JPG

    • PNG

Process

Per account request, a Professional Services team member can start the automated Cognitive Image Recognition process on a case. Any keywords identified in a specific image will be maintained in a custom text field and automatically added to the search index. The keywords can be displayed as a field on the Browse screen and will also be visible from the Doc Viewer.

Browse screen field with filter:

Doc Viewer

Additional Information

To request this service or additional information, please contact professionalservices@lexbe.com