How recipient recognition works
Envoy Deliveries identifies delivery recipients and notifies them automatically—all when you take a photo. But how does this work?
The basis of this process is optical character recognition (OCR) technology. OCR analyzes images (like photos or scanned documents), and “looks” for shapes that resemble letters and numbers. When a character is recognized, it’s recorded digitally.
With Envoy Deliveries, the recipient recognition process has three main steps: You take a clear photo of the shipping label. Your photo gets uploaded to an image transcriber. This image transcriber provides a digital transcription of the shipping label. Our system looks for matches between that digital transcription and recipients listed in your employee directory.
Step 1: Photo-taking
Recipient name recognition will be most successful with a clear, high-quality photo. Here are a few tips on getting the best photo.
Position the recipient name in the square outline
Take the photo in portrait orientation (not landscape)
Contrast is key:
Take the photos in a well-lit area
Use flash if it is dark where you record deliveries
Don’t write on the shipping label
If highlighting names, do so with a light color (yellow, bright green)
Take in-focus photos: Blurry or “too close” photos result in transcription errors
Step 2: Photo upload and processing
Once you snap the photo, your job is done. It automatically uploads to our image transcriber and the OCR process begins. After processing, the image transcriber provides Deliveries with the transcription text that corresponds with your photo.
This transcription text is broken into two sections. The full transcription text corresponds with the full photo. The highlighted transcription text corresponds with the portion of the photo that appeared within the square outline you saw on the Add deliveries screen.
Step 3: Recipient matching
In this step, Envoy Deliveries will compare the names in your directory with the transcription text block created in step 2.
Built-in troubleshooting
It’s important to note here that OCR technology—while quite advanced—isn’t always perfect. The image transcriber (which is a computer) can get “confused” by letters that look like numbers, odd spacing or punctuation, and more.
To combat these common issues, Envoy Deliveries does not simply look for an exact match between the transcription and your directory. Deliveries systematically tries many different techniques to find a recipient match before “giving up” on any delivery.
Rule: Add a to
• How it works: Deliveries adds the word to in front of all full names in directory, then looks for these phrases in the transcription.
• Name in directory: Alexa Beckfield
• What Deliveries looks for in transcription: To Alexa BeckfieldRule: As-is name
• How it works: Deliveries looks for the recipient name in the transcription (with and without accented characters, if applicable).
• Name in directory: Alexa Beckfield
• What Deliveries looks for in transcription: Alexa BeckfieldRule: Replace common transcription errors
• How it works: Deliveries replaces common mistrancriptions like B/8, 0/O, etc.Rule: Skip middle names
• How it works: Deliveries removes any middle names that appear in the directory, then looks for these revised names in the transcription.
• Name in directory: Alexa Jane Beckfield
• What Deliveries looks for in transcription: Alexa BeckfieldRule: Last name, first name
• How it works: Deliveries takes any names it “sees” in the transcript, reverses the order, and adds a comma to mimic this common format.
• Name in directory: Alexa Beckfield
• What Deliveries looks for in transcription: Beckfield, AlexaRule: Skip word
• How it works: Think of this rule as the opposite of the previous one. Deliveries looks for partial matches between names in the directory and the transcription, so if a shipping label showed a recipient’s middle name and the directory does not, it would still result in a match.
• Name in directory: Alexa Beckfield.
• What will match in transcription: Alexa XXXX BeckfieldRule: Strip spaces and punctuation
• How it works: Spaces and punctuation are often mistranscribed, so this rule combats extraneous punctuation.Rule: Abbreviated first name search
• How it works: Deliveries finds any possible last names in the transcription and “reads” the beginning of the associated first name. If at least the first three letters of the perceived first name match with any names in the directory, you will get a match.
• Name in directory: Alexa Beckfield.
• What Deliveries looks for in transcription: Ale… BeckfieldRule: Given name static listHow it works: Deliveries looks across the transcription and the directory, replacing given names with their common shortened forms and vice versa.
• Name in directory (example 1): Alexa Beckfield
• What Deliveries looks for in the transcription: Alexandra BeckfieldName in directory (example 2): Alexandra Beckfield. What Deliveries looks for in the transcription: Alexa BeckfieldRule: Levenshtein or at a distance search
• How it works: This search helps with spelling or transcription errors, since it compares the transcription to names in the directory for a “close enough” match. • • Name in directory: Alexa Beckfield
• What will match in transcription: Alexa Beckfeild, Aleea Beckfield, Alexa Beckfield, etc.
Note: This rule only works on two-word namesRule: Chomping
• How it works: Deliveries cuts off letters from the beginning and end of names in directory, and looks for matches in the transcription.
• Name in directory: Alexa Beckfield
• What will match in transcription: lexa Beckfield, exa Beckfield, Alexa Beckfiel, Alexa Beckfie, etc.
Nickname matching
If none of the rules above result in an automatic match to an employee in the directory, then Deliveries moves on to your nicknames.
In this example, employee Alexa Beckfield is the office manager. She has been set up with the nickname Facilities. If Deliveries finds the word facilities in the transcription, Alexa will be notified.
Next steps
As discussed above, Envoy Deliveries receives the full transcription text and the highlighted portion. The rules run first on the highlighted portion. If no match is found within the highlighted portion, all of the rules are run again on the full transcription text.
If no matches are found after running the full transcription text, Deliveries returns the Unidentified recipient notice.