top of page

Zebra Aurora™ Deep Learning OCR Tool



Reliable, Accurate Reads

Without Training

Zebra Aurora Interface - Deep Learning OCR job setup


Zebra Aurora software is a unified platform that gives end users of all skill sets control over all Zebrafixed industrial scanners and machine vision smart cameras, simplifying management of enterprise-widemanufacturing and logistics automation solutions. Among the software’s many capabilities is opticalcharacter recognition (OCR), which automatically extracts information from images.Oftentimes, training and retraining consumes a significant amount of time during OCR setup. Companiestoday need machine vision products that can dynamically adjust to such scenarios and deliver a reliableread. Traditional OCR applications may fall short when working with tough to read or low contrast characterson confusing backgrounds. With the new Deep Learning OCR tool, font training becomes redundant, as itdelivers reliable, accurate reads without having to train numerous different texts or fonts. The tool leveragesstate-of-the-art techniques that allow novices to quickly and easily set up highly accurate text recognition andcharacter reading applications.














Zebra Aurora™ Deep Learning OCR Tool

Approachability for End Users


End users unfamiliar with OCR technology or general machine vision may encounter challenges in systemsetup. Challenges that can impact success include how a code is printed, the type of surface its appliedon, the way the code is illuminated, and the way its captured by the camera. The Deep Learning OCR toolremoves these roadblocks. The approachable, easy-to-deploy machine vision system requires no coding orprogramming expertise. This allows companies to quickly install new deep learning-based OCR applicationsthat could previously not be automated.


Zebra Aurora™ Deep Learning OCR Tool

Swift Setup for Systems Integrators


For systems integrators, Zebra Aurora’s Deep Learning OCR tool saves significant setup and deploymenttime. The software requires no font training. Most non-deep learning OCR technologies require samples ofall potential OCR fonts — and all possible variations of a given font — and subsequent time spent training thesoftware on them. Integrators might spend months setting up an OCR font because the system has not yetencountered a particular variant of letter, such as one that is rotated or skewed or stretched or compresseddue to variations in line speed. Instead, the integrator can setup the scanner or camera and turn the intuitiveinterface over to a control engineer or another plant floor operator to begin using the deep learning–poweredOCR tool.


As Simple as Drawing a Box

Zero Training for Faster Deployment


















Zebra’s DLB OCR Tool can handle a great amount of process variation. The robust and powerful algorithms ensure maximum readability and throughput.

Manufacturing and logistics environments need machine vision systems that offer flexibility and ease of use. Requirements can quickly change, or new parts can be introduced. Streamlining the process of adjusting an automation system helps make flexibility and ease of use possible. Zebra Aurora’s Deep Learning OCR offers a pretrained model that allows users to deploy the tool directly on a camera, removing the time-intensive training portion of deep learning development entirely, including the training of new fonts or texts.


East of Use


Manufacturing and logistics environments need machine vision systems that offer flexibility and ease of use. Requirements can quickly change, or new parts can be introduced. Streamlining the process of adjusting an automation system helps make flexibility and ease of use possible. Zebra Aurora’s Deep Learning OCR offers a pretrained model that allows users to deploy the tool directly on a camera, removing the time-intensive training portion of deep learning development entirely, including the training of new fonts or texts


Confident Code Reading


With the Deep Learning OCR tool, end users can set minimum confidence scores to ensure a reliable read. In addition, users can look at the overall confidence of a given string and then drill down to the character-to-character quality to ensure the algorithm is reading correctly and that the highest-confidence reads are being delivered and then make adjustments as needed.


Choose From a Range Machine Vision Toolsets


With the new OCR tool, end users can draw a box around an area of interest for fixturing, which also teaches the algorithm the fonts, and capture textual information from the image. End users also can set thresholds that will help differentiate between different character strings, such as mathematical, alphanumeric size, or spacing. In addition, decoder performance packages can be added to any FS fixed industrial scanner or VS smart camera once. End users also have a range of machine vision tool sets available to choose from.



Tool

Description

Sensor

Standard

Object Locate

Find high-contrast features

Pixel Counter

Count the number of pixels in a designated range

Brightness

Return the average brightness of an area

Contrast

Return the average contrast of an area

Edge Tool

Find edges for fixturing and presence/absence

Distance Tool

Measure the distance between two tools previously created

Advanced Pattern

Find challenging features with variant lights, scaling, occlusions, etc.

Blob

Find, sort, and count areas with similar pixels

1D/2D/DPM

Read 1D, 2D, and DPM barcodes

Find Circle

Find and measure circles

Caliper Tool

Measure the distance by finding two edges

Filters

Enhance image quality for robust inspections

OCV/Flaw

Find residual or missing pixels from a trained shape


Power-Flo - Aurora Deep Learning OCR Brochure
.pdf
Download PDF • 2.50MB

Comentarios


bottom of page