Wednesday, February 24, 2021

RootsTech Connect 2021 - Day 1 - Innovations

After all the waiting, class preparation, session development, meetings, and everything else over the last couple months we are finally ready to start RootsTech Connect 2021. This was an incredible undertaking by the RootsTech team to get everything set up in such a short time and to help all of us with deadlines and questions to get to a point where this conference is today. And after all the work, we were able to experience the opening of the Expo Hall tonight and get ready for the first set of talks. 

One of my favorite parts of RootsTech is talking to the innovators and exhibitors. This year there are 14 Innovators at RootsTech. Some of the innovators, such as Adobe and MyHeritage, are showing off their photo restoration and enhancement capabilities. Adobe has their Photoshop, Lightroom, and Spark software on display. While MyHeritage was showing off their colorization and photo enhancement tools that they introduced in 2020. MyHeritage has had over 30 million photos enhanced using these applications. Adobe and MyHeritage are on total opposite ends of the spectrum for photo restoration and enhancement. MyHeritage allows users to upload their photos to the website for free and does a quick enhancement with no user based controls. MyHeritage's products are impressive and easy to do. Adobe is a more complex process which may require the user to manipulate various controls to get a specific outcome. Of course the user has more control with the Adobe products. Also, Spark provides ways to share your family photos on the web or in social media.

Another area of RootsTech that I like is the Demo Theater. This is where sponsors and exhibitors talk about their new developments, helpful products, and many other topics.

For tonight, my favorite innovation is FamilySearch's demonstration of their computer assisted indexing which is designed to be able to read handwriting in multiple languages. I tried it out with a marriage record written in German. It was pretty good but still had some errors. Overall, I think it did well enough to get a good start on the translation considering the quality of the document it had to start with. You can see the demo of this technology in the RootsTech Sessions at Computer Assisted Indexing. Then you can go to the website and see how well it will transcribe your handwritten documents. Here is an example of what it did with my German marriage record. The colors indicate people, places, and events.


The program is capable of reading handwriting in a variety of languages including German, Cyrillic, Arabic, English, and many others. So this should help you with a variety of documents.
 
FamilySearch also provides a demonstration video that shows a more difficult example where the computer is able to recognize the handwriting and extract the information into a tree showing the relationships included in the document. You can view the video of this demonstration on their session page. I'm sorry about the poor quality screenshots but they are clipped from the YouTube video.



Previously, the team had been working on improving the computer's ability to recognize typed document. Many of those records are now available on FamilySearch. You can see an example of an indexed record here. This work was connected to the Zoning Project which I participated in several years ago. The Zoning Project identified parts of newspapers which included important genealogical information such as birth, death and marriages and marked them so the computer would be able to recognize individual pieces of information. I did a blog post on the Zoning Project back in 2017 which you can read here.

Why is this innovation important? Once the computer has sufficiently learned this process, it will be able to utilize cloud based processing to index more records in one day than a person could in their entire lifetime. Currently, there is a quality threshold that has to be met before these computer assisted indexing projects are sent to the FamilySearch collections. Here is an example of a Spanish language church record that the computer indexed. Soon, you will begin to notice more records being indexed by computer. You will be able to identify those records since they have a blue box stating that the record was indexed by a computer. These records also have the capability for the user to edit mistakes.

Not only will this increase the number of searchable records but it will also allow you to request computer assisted indexing on demand for unindexed records. That means that any unindexed images (80% of all images are currently unindexed) that you find can be indexed on demand. There will be a button on the screen that will index the record and provide the record for you to review and edit. Eventually, as images are acquired they will be indexed y computers and then put online within a short period of time.

Innovations such as these will greatly increase our capability to be successful in our research.

No comments: