I used Mobile Vision API to detect US Driver Licenses. I am able to scan the DL PDF417 barcodes and I see that driverLicense object is set within Barcode object. However for some states (Eg. FL, NC) the driverLicense object is NULL.

When I checked the valueFormat in Barcode object it is set to 12 (Barcode.DRIVER_LICENSE) for working scenario and set to 7 (Barcode.TEXT) for non-working scenario. Could you confirm whether this is an issue with BarcodeScanner APIs? Sample code is as follows.

The DL/ID parser is fairly simple and doesn't work perfectly on IDs from all states, unfortunately. There might be some jurisdiction-specific fields that are tripping it up, or they might not be using a fully spec-compliant encoding.

We'd like to improve it -- if you'd like to help, the anonymized raw text of the non-working cards would be helpful in diagnosing the problem. However, if you aren't comfortable sharing it (or just aren't sure how to completely remove all identifying information) that's okay too!

If you can't share the raw values, it would be helpful to know where the first occurrence of 'DL' appears in the values that won't parse. Normally, it would be at index 21 (e.g. '@n rANSI 636045030002DL').

Sure, here is the raw data from the scanner. I used the sample license from internet for FL state.
03-21 14:41:49.422 4238 5774 D BarcodeScannerActivity: Barcode Format: 2048
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: Barcode Raw: @
ANSI 6360100102DL00390201ZF02400052DLDAASAMPLE,IDENTIFICATION,CARD
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DAG2900 APALACHEE PKWY
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DAITALLAHASSEE
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DAJFL
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DAK32399-6552
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DAQS514403827471
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DARI
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DAS
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DAT
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DBA20110707
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DBB19820707
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DBC2
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DBD20070502
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: DBHN
ZFZFADUPLICATE: 20070504 5774 D BarcodeScannerActivity: DAU507
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: ZFB
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: ZFCQ010705040016
03-21 14:41:49.429 4238 5774 D BarcodeScannerActivity: ZFD

Sample FL DL.PDF
Attached the sample license card.

I got the NC state info and removed the personal info. Let me know whether this helps. This is from a valid NC state DL.

@
� AAMVA36004001DL00280208DLDABLASTNAME
DACFIRSTNAME
DADMIDDLE MIDDLE MIDDLE
DAE
DAL123 SAMPLE ST
DAM
DANCHARLOTTE
DAONC
DAP12345-6789
DAQ99999999
DARC
DASNone
DATNone
DAV6-04
DAYBLU
DAZBRO
DBA11-28-2019
DBB04-01-1984
DBCM
DBD06-11-2012
DBHY

Here is the NJ state DL which is working fine, 03-21 17:53:57.419 28002 28431 D BarcodeScannerActivity: Barcode Format: 2048
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: Barcode Raw: @
ANSI 636036040002DL00410318ZN03590153DLDAQK00000000000000
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DCSLASTNAME
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DDEU
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DACFIRSTNAME
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DDFU
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DAD
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DDGU
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DCAD
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DCBNONE
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DCDNONE
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DBD02222014
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DBB01011900
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DBA03192017
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DBC1
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DAU064 in
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DAYBLK
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DAG107 BERMUDA DR
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DAINESHANIC STATI
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DAJNJ
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DAK088534286
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DCFED201405300000212
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DCGUSA
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DDAB
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: DDB07232010
ZNZNAFN7:53:57.420 28002 28431 D BarcodeScannerActivity: DDDT
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: ZNB19.88
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: ZNCRENC
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: ZNETEMP IMMGR STATUS
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: ZNFBLK
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: ZNH88:igJvvXLJs8EWsvn8in7RvzDTmCOYM3CSgdgdgdfFgg/u0SdDaYt2IXGALST10ufQM8WmBSciv60GNuORlAh82cQ;0291;
03-21 17:53:57.420 28002 28431 D BarcodeScannerActivity: Barcode Obj:com.google.android.gms.vision.barcode.Barcode$DriverLicense@80230b

Thanks, this is very helpful (though in the future, it would be great to get an escaped literal string that preserves special characters, e.g. '@n rANSI [...]r').

The FL and NC licenses both violate the AAMVA spec in a couple ways, so we're going to make our parser more lenient for future releases.

Thanks for the update. Could you confirm the timeline for the release with improved parser?

We don't have a timeline to announce, but I expect a new release within a couple of months.

Oregon license sticker (the owner moved and received a sticker from the DMV to replace the barcode with new street address information).
There are lots of trailing spaces for some reason and the ANSI/DL code is missing.

Wow. That doesn't even attempt to follow the specs -- unfortunately, our parser won't recognize that text as DL/ID data without the leading '@n r(ANSI|AAMVA)'. It also doesn't contain the required directory information.

It's pretty bad. I'm just going to write a super lenient fallback for the case where despite best efforts the BarcodeDecoder can only return Barcode.TEXT. Might need to do a switch or something, based on what comes out of the DAJ field. I hope Oregon gets its things in order soon, but even if they fixed it tomorrow people would still be carrying around licenses like this through 2021.

Actually, please disregard the earlier sample I posted; I'm not sure why, but escaping the string correctly and dumping it on a single line made the prefices and suffices I was expecting show up.

These samples are also from Oregon and don't register as driver's licenses, even though they do have the @n rANSI prefix. They're malformed in other ways, of course...
(1234567 is a placeholder for their ID number)

@nNrANSI 6360290102DL00390285ZO03240027DLDAQ1234567 nDAALAST, FIRST MIDDLE nDAG nDAL5555 NUMBERED AVE nDAIPORTLAND nDAJORnDAK97215 nDARC I nDASBD nDAT nDAU009nDAW135nDBAYYYYMMDDnDBBYYYYMMDDnDBC1nDBDYYYYMMDDnrZOZOAFIRST LICENSED YYYYMMDDr

@n�rANSI 6360290102DL00390192ZO02310031DLDAQ1234567nDAASECOND LAST, FIRST MnDAGnDAL55555 NAMED DR APT 101nDAIBEAVERTONnDAJORnDAK97006 nDARC nDASBD nDAT nDAU509nDAW135nDBAYYYYMMDDnDBBYYYYMMDDnDBC1nDBDYYYYMMDDnZOZOAFIRST LICENSED MM-DD-YYYYn

(Yes, the format of the date ZOZO field is different between the two.)

Thanks for the examples!

However, it appears that the process of redacting personal information may have changed the length of some fields (our updated parser fails on the snippets you provided). The large block of numbers in DL/ID data is a bunch of offsets, which means you must take care not to add/remove any characters when sanitizing it.

Ah, my mistake. These should match the original byte positions:

@nNrANSI 6360290102DL00390285ZO03240027DLDAQ1234567 nDAALASTN, FIRST MIDDLE nDAG nDAL5555 NE NNTH AVE nDAIPORTLAND nDAJORnDAK97215 nDARC I nDASBD nDAT nDAU009nDAW135nDBAYYYYMMDDnDBBYYYYMMDDnDBC1nDBDYYYYMMDDnrZOZOAFIRST LICENSED YYYYMMDDr

@n�rANSI 6360290102DL00390192ZO02310031DLDAQ1234567nDAALASTN, FIRST MIDDLEnDAGnDAL55555 NE ANOTHER DR APT 101nDAIBEAVERTONnDAJORnDAK97006 nDARC nDASBD nDAT nDAU509nDAW135nDBAYYYYMMDDnDBBYYYYMMDDnDBC1nDBDYYYYMMDDnZOZOAFIRST LICENSED MM-DD-YYYYn

Unfortunately, both of those examples are badly broken.

The first claims to contain a DL block starting at offset 39 with length 285 ('DL00390285') that overruns the entire string. The second declares a block ('DL00390192') that extends partway into the next ZO block.

Doesn't detect barcode in this sample image which contains data for a NY passport.

Tried the barcode reader sample on a Nexus 5.

Thanks for the sample. The raw contents ('@ nnnANSI') are unfortunately wildly incompatible with the specification, which says entries begin with '@n rANSI'.

However, I suspect that many sample drivers licenses are produced manually and are full of errors that don't occur 'in the wild'. Real New York licenses decode fine with our parser.

Thanks for the explanation @pchx

The latest version of our Barcode API (part of Google Play Services 10.0) makes the DL/ID parser more lenient; it should handle a wider range of licenses now (there may still be some remaining edge cases, however).

Cool @pchx. Let me test it out this week. Thanks for the heads up.

Hello @pchx. I was wondering when does length start, i understand that it starts before 0180 and ends at the bottom?
@
�
ANSI 6360350201DL00290180DLDAASAZZLAUZKAS,JUZZAS
DAQS24242043364
DBA20171223
DBB19461213
DAG30 LILWE AVE
DAIPOX LAKE
DAJIL
DAK600600000
DARD
DASB
DAT*****
DBD20150115
DBCM
DAU509
DAW120
DAYBRO

Not sure if this thread is still alive but Im trying to parse some drivers license at work. The issue Im running into is that the data from the scanner is coming back as one line with no return characters to separate the data fields. Anyone else run into this issue?

Can someone fill me in on the exact format including the /n and /r? Also, what is the diamond character with the questionmark in it? Anyone know?

@emottau I would suggest to refer to AAMVA DL spec (Table D.1 — 2D symbols header format). Also, if you see any issue in parsing please open new issue as this is closed.

Can someone fill me in on the exact format including the /n and /r? Also, what is the diamond character with the questionmark in it? Anyone know?

/n is line feed that separates the individual data objects in the record. /r is carriage return that as far as I can tell helps locate beginning and end to the record. For example in the Arizona DL AAMVA 2D barcode the record is ended with (nnr) this is the only occurrence of this pattern so I can easly know when the record is completely read. The diamond is actual what's called a record separator 'rs'. When decoding the buffer it has no clue what it is. In javascript I was able to use it to denote the start of the record in an Arizona DL using this code: str.includes('u001er'+'ANSI') record separator, carriage return and 'ANSI' is the beginning of the record.

Not sure if this thread is still alive but Im trying to parse some drivers license at work. The issue Im running into is that the data from the scanner is coming back as one line with no return characters to separate the data fields. Anyone else run into this issue?

My guess it could be your bar code scanner is programmed to trim any spaces and hidden characters as it outputs . Try factory reset on the scanner.