Significantly higher demand of authority work due to the COVID-19 crisis – but capacity is limited. E.g. we have heard in the news that many Swiss companies have already registered short-time work for more than 1 million employees. This means that authorities have to deal with many different kind of forms and requests.
What it does
In our current MVP the tool extracts relevant information out of COVID-19 pre-registration form for short-time work (PDF format). This can be further extended to other forms.
How we built it
We use opensource technologies for the following activities:
- OCR: Converting scans to text
- Text recognition
- Information extraction
Challenges we ran into
The surrounding was challenging (virtual meetings, communication). From a technical point of view, we had issues with text recognition, not all content could be easily extracted.
Accomplishments that we're proud of
We have a running prototype in place :)
What we learned
We learned a lot about image recognition by using different open source tools.
What's next for The Form Reliever
- make the software production ready
- scale to other incoming ad-hoc forms during crisis
- extend to hand-written forms
Try It out
flask, formdetection, pandas, python, tesseract, textrecognition