Using optical recognition techniques in field data gathering?

Jim Simmons, Research Fellow

Gathering field data has been, for the most part, comprised of taking notes at the subject property on a pad of paper. Some appraisers have adopted some type of worksheet form, either pre–printed or self–designed, to help as a checklist of items needed to fulfill the data requirements of the report they will be completing.

The worksheet I use is illustrated in Fig 1 and is basically unchanged since the URAR was last revised. It is simply a quick cut and paste of the SITE and IMPROVEMENT sections with a homemade tack–on for use with FHA reports. This is then taped to the top of a blank legal sized page and copied. The bottom two thirds or so leave room for my sketch and notes. While not pretty, it works well for me.

Many have purchased a PDA and use Pocket TOTAL for their mobile data collection solution. I tried a PDA and found it worked well but had difficulty with the screen size. It is better suited for someone with younger eyes. Another suggestion has been to use a Tablet PC. For fieldwork you would want one rugged enough to handle the riggers of heavy–duty usage. I was able to use a Panasonic Toughbook for a few months some time ago and it's built like a tank. The screen size is large enough to handle a font size suitable for the reading glass crowd.

Both PDAs and Tablets pose the same downsides. Both (particularly the Toughbook) pose price issues and there is still a need to take field notes although both have a stylus input. Interior inspection could be handled easily by these units but once you include the exterior inspection into the mix the juggling act begins. Add a tape (and/or Disto) and a camera and well, there just aren't enough hands to effectively include another input device.

I'm thinking the advances in nanotechnology and mobile programming could overcome many of these deficiencies in the future but what about for the near term? Seeking alternatives to these methods of field collection, our challenge is to create a system that both acts as a checklist and possibly as an input mechanism to automate the transfer of field data to WinTOTAL form data, while eliminating the need for the appraiser to purchase any additional hardware.

I would propose a bi–level strategy. First, field worksheets could be generated by WinTOTAL that are unique to a specific appraiser, for a specific type form by utilizing a combination of fixed checkbox items and multiple choice checkbox list generated from the resident "quicklist" data fields. An appraiser would use the worksheet as a checklist when performing the site inspection. Upon returning to the office the worksheet would be used as a reference to enter the data into WinTOTAL.

Second, an interface could be developed for an appraiser with an XSite that would leverage the DirectFax system supplemented by available OMR and/or OCR technologies. The process would work something like this: Appraisal request is logged and a field worksheet form is printed. The top half should accommodate required entry fields plus digital coding to allow server software to identify, process and route the digital image. The bottom half of a legal sized form should leave enough space for a sketch and ample notes.

When the appraiser returns to the office (or even before if one had access to a fax machine) the field sheet is faxed to the processing server with a DirectFax number. Now the received fax is taken by the server program, OMR converts check marks to field data responses and possibly OCR converts text to limited fill–in responses. The resulting data and digitized worksheet is then delivered back to the appraiser's form for autoentry into the proper fields with an attached digital field–worksheet in the work file.

So what is OCR, OMR and how can we get from field to form? Briefly OMR stands for Optical Character Recognition. This technology has been used for many years to get scanned pages of typed text into editable documents, say in Microsoft Word. OMR is Optical Mark Recognition and is the driving engine behind those Scantron test cards used in school. Fill in those bubbles and a scanner unit automatically scores the results.

Another application is in the medical field. Medical aid technicians gather vital medical histories on patients being admitted for treatment. The fact information is scanned directly from data taker to the database reduces the probability of entry error by half. Of course, field data gathering for an appraiser is not a matter of life or death. It is much more important than that. According to research there are programs that allow recognition of X marks, slashes and other more exotic variants using a process called "fuzzy logic."

But enough "techno–speak," on with the process.

By matching the sequence of each field with the corresponding response the resulting checkmark could be translated to the appropriate word description in the appraisal form. This would be unique to the individual appraiser and therefore produce a custom field checklist for each appraiser for each form type. The top half of Fig 2 is a screen shot that illustrates what the dropdown list looks like for a typical field. The bottom half of Fig 2 shows a possible look of the same field converted to a field worksheet entry checklist.

To obtain the form result of Stucco/Average in this instance the appraiser would mark the STUCCO box under the Exterior Walls column and AVERAGE under the Condition column. The additional boxes adjacent to the last box under VINYL can accommodate an OCR entry of up to six characters.

I think the first phase should just include the unique entries in the "quicklist" dropdown list for each relative field. The next step should be the addition of a database containing the last 10 unique responses even though they may not be included in the "quicklist." They would still tend to be ordered by frequency of use so the most often used words would be toward the top of the list.

This is not meant to replace any other field data gathering tools, but as another method of getting data from the field to the form in the least amount of effort on the appraisers' part. Send feedback to let us know what you think.