Steve Kelley
For those of us who have been waiting rather impatiently for Microsoft’s Seeing AI app to come to Android, the wait is over—sort of. Google’s Lookout App made its debut last year, and originally was only available on the Pixel phone and some select others. If you gave the earlier versions of the app a try and abandoned it for its early limitations, it’s time for another look.
Google’s Lookout App is now a more polished app that will run, according to the system requirements outlined in the Support documentation, on Android 6.0 (Marshmallow) or above. For this review it was running on a Galaxy S9 with Android 10. Download the free Lookout App from the Google Play Store. The first time the app opens, it asks for your Google account or prompts you to create one. This is followed by the generic request for permission to share data use. To use the app, of course, you grant permission.
Opening the App
By default, the app is in Text Mode when it opens. Text Mode is just one of the five modes available in Lookout. The various modes appear along the bottom of the screen, and include: Explore (listed as beta); Food Labels; Text; Documents; and Currency. On the top portion of the screen is the Change Language button, which offers the option of selecting any of 22 languages. To the right of this button is the Account Menu, which contains options for Settings, Help and Feedback, and Contact Info. Help and Feedback is a great place to get started with various support articles that explain the various modes and settings options.
Text Mode
Text Mode is designed to read any text the camera detects, whether it’s on a product label, newspaper, or computer screen. Text Mode is quick, and moving the camera around will jump from one item to the next, which can be confusing. Holding the camera steady and gradually moving it across the text proved the best way to read without a lot of jumping around.
This mode is great for reading short bits of text like an address on an envelope, and for scanning a pamphlet, business card, or medicine bottle. The default setting for the app is to use the device’s flashlight in low-light settings. For nearby observers, this can look like a strobe light as it goes off and on to light up the text. The flashlight can be disabled in the Settings menu for more discreet reading, and Lookout does pretty well in dim light, without the flashlight.
Reading in Text Mode is responsive, quick, and pretty accurate. The text-to-speech voice pitch and rate can be changed from the Settings menu.
Document Mode
For longer text recognition, the Document Mode will scan a page of text and convert it to digital text, accessible to TalkBack or Select to Speak. Document Mode offers nearly continuous hints to align the document properly before snapping the picture. You might hear, “Too close, move the device away,” or “Move device left,” etc. until the document is within the camera frame. Once framed, Lookout will report, “Hold still,” or “Try taking a snapshot.” The camera automatically takes a picture following “Hold still.” When prompted to take a snapshot, press or double tap the Take Snapshot button to manually take the picture.
Text recognition is super-fast and seems very accurate. During the review, Airplane Mode was turned on while using Document Mode, and this had no impact on text recognition, so optical character recognition (OCR) processing must be done on the device rather than in the cloud. In addition to being fast, it makes it so much more convenient to have text recognition available all the time, with or without internet access.
Once a document is processed, the text can be read, copied and pasted, or shared using the Share Button in the top right of the display. Document text can be shared with other apps or email. What is missing here though, as with Seeing AI, is the ability to scan multiple docs and save them as a file for later use, like you can with an app like KNFB Reader. As long as the Lookout App remains open, you can open the Recents button, on the right side of the screen just above the Modes list, to open any recent scans. This list disappears once the app closes.
Explore Mode (Beta)
Explore Mode describes objects in the environment as the camera is moved around. In the kitchen, for example, it quickly identified cabinets, the oven, a microwave, coffee cup, tableware, window, door, etc. It identified the cat as a dog, and failed to ID the fridge and other smaller appliances. One of the more impressive identifications was identifying a backpack as a bag, and then reading the text on a business card in a clear plastic sleeve attached to the bag. At the time, the bag was about three feet away, in dim light.
The Explore Mode, like Text and Document, functioned in Airplane Mode, so this too must be working off device-based object identification. Although in beta, it certainly appears useful for identifying many of the major features in the immediate surroundings, such as the door, window, and furniture. It may be premature, however, to delete your Be My Eyes App or any other app using a human to identify objects or describe details in the environment around you.
Food Labels Mode
The first time Food Labels Mode was opened, an additional data file was downloaded, which took a couple minutes, and was specific to the United States. The US is just one of 20 countries available to select from. The onboard support article indicates that data files are not available for all countries. It also indicates these data files enable offline barcode recognition.
Like Document and Text modes, Food Label Mode is very responsive and fast. One of the most impressive features was the ability of the camera to locate the barcode on a wide variety of products. Following the directions provided in the support article for Food Labels, I slowly rotated products in front of the camera. In most cases, identification occurred quickly. The flashlight is used for dim lighting as in a kitchen cupboard or pantry. One of the best features in Food Label mode is the way that barcodes and text recognition are incorporated, so if the barcode is not immediately recognized, the text on the product label may be recognized first and read. Once the barcode is identified, it often provides more details about the product than what is read from the label. In addition to the quick response the on-device data provides, this feature will make Food Label Mode really handy while shopping in a grocery store, where access to the internet might be spotty.
Currency Mode
Like the other modes, Currency Mode takes its settings by default from the phone, so if you are in the US, the currency will be U.S. Currency. At least 20 other countries are listed in settings, so it's simple to change currencies. Like the Text and Food Label modes, Currency Mode is really fast and accurate. It identified various bills in a variety of conditions—folded up, upside down, etc., and in each case the response was accurate and nearly instantaneous. Because identification happens on-device rather than relying on cloud service, this can be a really handy feature anywhere.
Conclusion
Like Seeing AI for the iOS devices, Google’s Lookout for Android is a game changer. With text and barcode recognition on the device, processing seems nearly instantaneous in all five modes. Text, barcode, and currency accuracy is quite reliable, even with some of the stylized text that appears on package labeling. If you’ve been waiting for Microsoft to migrate Seeing AI to your Android, just download Google’s Lookout on the Play Store——you may not need Seeing AI after all!
This article is made possible in part by generous funding from the James H. and Alice Teubert Charitable Trust, Huntington, West Virginia.
Related Articles:
- Supersense from Mediate: A First Look at a New Object and Text Recognition Mobile App by Bill Holton
- A Comparison of Speak! and Envision AI, Two Text and Object Recognition Apps for Android by Steve Kelley
More by this author:
