Thanks for trying it out and sharing your findings. Really appreciate it.
- We're looking into the "Select from Server" not working and will keep you posted.
- The TB tool session expiring is being handled with a suitable error message saying "Session expired". This will be functional in the upcoming TB version [mostly bug fixes from the present version].
- Fair point regarding the illumination.
- Handwriting recognition is not supported for Rev A. Please put yourself out of misery trying it out for now. We're investigating this for a future release. Will keep you posted.
- We can definitely walk through supported and unsupported features in the next call.
If it's not too much trouble, please also attach images of the forms you're trying to read. Would be useful for us to train our software, while functioning as samples for others at the same time.
I'd second Horsts' point vis illumination - also it is worth noting that if you set illumination to Auto, the LED remains on post-capture.
I'm still playing around in the shallow end, but I have noticed some consistent errors, even in the sample documents: for example the Postal/T&L doc regularly mis-decodes the "1" in the From address postcode field as an "I", and similarly adds a space character to the Delivery address postcode.
Since my first stop from here is to play with customer labels and connotes, do we have any expectation as to how the software may work with dot-matrix printed charcters, particularly in dealing with poor contrast?
Out the gate I must say I am super impressed though! Well done to the team!
Hi Horst, Jon,
Glad to see that you guys are able to get everything up and running. Looking forward to hearing more pain points so that we can correct them as we go along.
-The "Select from Server" problem occurring has been identified and corrected on our end. In our next update, this issue will be rectified.
-We have also confirmed the issue in which the flash stays on when viewing the results. You will see this fix in the next update as well.
-We have noticed the OCR issues you are experiencing, and will work with the appropriate team to get them resolved. However, keep in mind that you will not always get 100% correct results, due to the nature of OCR recognition and the quality of the image/text, which is why a visual check is recommended. We try to alleviate the trouble of this process by flagging low-confidence regions so they are easy to spot, but some situations may occur where a high-confidence region contains errors.
To my knowledge, dox-matrix fonts are supported; however poor contrast will make it difficult to extract correct results.
Thanks again for giving this a run-through!