Not all that long ago, separate apps were needed for text recognition, currency identification, bar code reading, and color identification. Seeing AI from Microsoft and Envision AI from Envision Technologies each offer a multi-purpose app that claims to perform many of these functions, and more. Both apps are available in many countries throughout the world.

For this article, an iPhone X was used to evaluate both apps; each item used was scanned three times to insure consistency.

Microsoft Seeing AI

Once the Seeing AI app is installed, there will be a notification asking to allow Seeing AI to use your phone's camera. A brief tutorial appears on the screen after permission is granted.

Seeing AI's main screen has four options: Menu, Quick Help, Pause Announcements, and Channel. The best way to navigate the home screen is to flick left or right with one finger.

The Menu Option

The first choice in the Menu is Browse Photos. Selecting this option will cause Seeing AI to ask if it can access your photos. If access is allowed, your photos will be displayed.

The next Menu choice is Face Recognition. In order for Seeing AI to recognize a face, the person must first be photographed and their name put into the database. The Face Recognition menu option explains how to do this.

The third Menu choice is Help. This section is easy to navigate by headings, and each feature is clearly described.

The fourth choice under the Menu option is Feedback. Selecting this option opens an e-mail message for you to send feedback about the app to the developers.

The fifth choice in the Menu is Settings, which provides opportunities for customization including setting voice and speech rate, which currency to recognize, setting up 3D touch shortcuts, and choosing whether the camera automatically uses a flash when necessary.

The final Menu choice is About. This section gives the current version number along with the fine print like the terms of use and the privacy policy.

The Home Screen

The Quick Help feature is context dependent, which means the options will change based on which feature is being used. For example, if Short Text is chosen, the information in Quick Help will describe how to use this feature and give an option to play a video on how to use it.

The "Pause Announcements" button is a toggle. When the button is active, VoiceOver will still speak.

The final item on the Home screen is Channel. This may sound confusing, but Channel indicates which feature is currently active. Think of it as different TV channels. Channels include Short Text, Document, Product, and Currency. Options such as Color and Scene will have the word Preview next to them. These features are available but still in development. Swipe up or down to get to a specific channel, or, if your device supports 3D touch, you can pick up to four channels to access with the feature.

Envision Technologies Envision AI

This app offers a free 14-day trial. After that, there is no cost for up to 10 actions per month. Subscription plans start at $4.99 per month.

The first time the app is used, there is a welcome screen and a "Log In" button. Activating this button brings you to a new page where you have the option to log in with Google, Facebook, or your e-mail address. Once you're signed in, Envision will ask for several permissions to access location, camera, and photos. A welcome email will also be sent.

Screen Layout

There are three tabs at the bottom of the main screen: Text, General, and Help. By default, when the app is first launched, the Text tab is selected. The easiest way to navigate the Home screen is with a one-finger flick.

With the Text tab selected, there are four options: Magnifier, Start Reading Instantly, Read Handwritten Text, and Read Document.

If the General tab is selected, the options are Describe Scene, Detect Colors, Scan Bar Code, and Teach Envision.

When the Help tab is selected, a screen loads with a help section and below it is the Settings section. The Help section has three options: Read Tutorials, Give Feedback, and Request a Call. When Read Tutorials is activated, the next page will require you to choose a language. This must be done every time. You can navigate the different tutorials using the headings option in the VoiceOver Rotor. The tutorials are clearly worded.

The first option in the Settings section is a toggle labeled "Offline text recognition, faster recognition option for languages based on Latin scripts." By default, this feature is off. The Speech option is next. Here is where speech parameters can be adjusted. Within the speech settings is a button labeled Automatic Language Detection. If you are only using one language, keep this button off. The next setting is for Color Detection. There are two options, Standard 30 and Descriptive 930. The second option is on by default.

Below the color settings is information about your account. There are options to Share with Friends and Write a Review on the App Store. The final option is About Envision.

Evaluation

I used the same lighting conditions for all tasks.

Short Text

In Seeing AI, hold the phone over the item and once text is found, the app will read it. The app will continue to look for text, which can be useful if looking for a room number or address.

With Envision AI, select the "Start Reading Instantly" button and hold the phone over the item. Envision will beep when it locates text. When finished, tap the "Stop Reading Instantly" button.

I had both apps read my computer screen, the label on a bottle of wine, and two store coupons. Both apps did well. Envision AI had slightly faster text recognition and did significantly better than Seeing AI with the wine bottle.

Documents

The easiest way to scan a document page with Seeing AI is to put the phone on the document with the camera in the middle. Gradually lift the phone up and the app will say which edges are not visible. Slowly lift the phone up until you hear "hold steady." This means that the document is going to be photographed in about two seconds. After the document is processed, it can be read using VoiceOver navigation or by activating the "Play" button at the bottom left of the screen. Additional options at the bottom of the screen are Stop, Increase Font Size, Decrease Font Size, and Share.

The set-up procedure is the same for Envision AI. Put the camera in the middle of the page and lift the phone up gradually. When the page is in focus, the app will say, "hold steady." It says this at the same time the shutter clicks. Once the document is processed, it can be read with VoiceOver commands or by activating the Play option at the bottom left of the screen. You'll find an Export Text option next to the Play option.

Envision AI lets you scan multiple pages in a document and then process them. Double-tap and hold on the "Document button". The app will ask if you want to read multiple pages. After the last page is photographed, activate the "Done" button to start processing.

Three different documents were used to test this feature: a letter, a financial report, and a printout of the USDA's food safety information. Seeing AI was more accurate with each of the documents, though Envision AI's results were still very good and the ability to scan multiple pages at once is very useful.

Reading Handwritten Text

Both apps indicate this feature is for reading short notes like the ones on sticky notes. They operate the same way except Seeing AI requires you to photograph the item manually while Envision AI takes the photo automatically.

Seeing AI read enough of the samples to let me get the basic idea of what was written. It even read my awful handwriting; however, Envision AI was not able to recognize handwriting on any of the test documents.

Currency Identification

At this time, only Seeing AI has a currency identifier. It read currency very quickly.

Product Identification

When using a bar code scanner, make sure the item to be scanned and the camera are positioned correctly. This is critical.

Seeing AI has a Product channel that is used for reading bar codes. Position the phone's camera so it faces the product and slowly move either the phone or the object. The app will play faster beeps as you get closer to the bar code. Once a bar code is identified, Seeing AI will speak the name of the item. If there is additional information about the item, there will be a More Info option on the screen, so reviewing the entire screen can be helpful. The same options that are on the screen when the Document channel is active appear on the More Info screen.

Envision AI's bar code scanner is in the General tab. Double tap the "Scan Bar Code" button. Follow the same instructions as above for scanning. You will hear beeps as you get closer to the bar code. Once it is identified, the app will speak the product's name. If more information is available, a More Information option will appear near the top of the screen.

A variety of test items were scanned, including cans, frozen foods, and product boxes. Both scanners worked equally well in these tests.

Color Identification

When attempting to identify colors, it is important to have good lighting. Seeing AI has a Color Preview channel that allows you to use this feature while it is in development. Hold the phone above or in front of the item to be described and the app will start speaking.

In Envision AI, position the camera above or in front of the item you want to scan. Activate the "Detect Colors" button from within the General tab. The app will start speaking.

I used six items for color identification including a red shirt, a pair of jeans and a light pink shirt. The app was tested in three different lighting conditions: two indoors and one in sunlight. Unfortunately, neither app did particularly well. Each got one color correct part of the time but not consistently.

Describe a Scene

These apps can take a photo of your surroundings and attempt to describe what is there. In Seeing AI, go to the Scene Preview channel. Once you are ready, activate the "Take Photo" button. The app will describe the surroundings. At the top of the screen is a "Close" button. Below the description are options to save and share the photo.

In Envision AI, first make sure your camera is pointed in exactly the right direction, then select the "Describe Scene" button from the General tab. Once the button is activated, the app will automatically take a photo and speak a description. There is an option to save the photo or, if you want to take another photo, you can reselect the "Describe Scene" button.

I photographed both indoors and outdoors. Both apps correctly indicated I had photographed my dog, a tree, a laptop, and my living room. However, Seeing AI correctly said I photographed my house while Envision AI kept saying, "looks like clear skies to me" and then gave me the temperature.

Face Recognition

In order for either app to recognize someone's face, the person must be photographed and added to the app's database.

In Seeing AI, go to the Menu at the top of the page and select Face Recognition. Select the "Add" button. By default, the front facing camera is selected so the person who you want to photograph can take selfies. You can switch to the back-facing camera if you prefer.

Once the photos are taken, the app can then recognize the person and describe what they are doing. Even if you do not put the person into the database, the app will give a brief description such as age, gender, and hair color.

In Envision AI, activate the "Teach Envision" button. There are options to photograph a person or an object. After taking the photos, activate the "Done" button. An edit box will open to enter a person's name or object. If the person or object is in a scene, the app will say their name or the object's name.

These apps worked equally well during these tasks, but neither worked perfectly.

Conclusion

Both apps offer excellent features. Fortunately, both Microsoft and Envision Technologies are dedicated to improving their products. Neither app is perfect, but they are very helpful, and both have excellent onboard help. Envision AI also has a Request a Call feature, which I used. I figured out the solution while I waited but still took the return call, which confirmed my solution. The call came after about an hour, which was impressive since it was Saturday afternoon. Also be aware that Envision AI is now available on Android through Google's Play Store.

Since Seeing AI is free and Envision AI offers a free trial. I recommend trying both, to find out which app works best for you. Both are powerful tools to increase independence in the home, at school, at work, or on the go. With either of these apps on your phone, you will not need to ask someone else to read you a menu or tell you the price, and you will never have to trust your bartender to give you the correct change!

Product Information

Product: Seeing AI
Developer: Microsoft
Price: Free
Product: Envision AI

Developer: Envision Technologies B. V.
Price: 14-day free trial with 10 free captures a month afterward. Subscription: 1 month $4.99, 6 months $24.99, Annual $33.99.

This article is made possible in part by generous funding from the James H. and Alice Teubert Charitable Trust, Huntington, West Virginia.

Comment on this article.

Related articles:

More by this author:

Author
Janet Ingber
Article Topic
Product Reviews and Guides