The Experiment
The Problem: Finding the Lost Sheep
There are over 6000 books in Dr Rick's Library. They are not all in the same location. An important
task is locating and documenting each of the books.
This library, or at least portions of it, has been all over the world. It started when I was a boy. I still have, somewhere, a book titled Quiet Boy which was written by Lela and Rufus Waltrip and illustrated by Theresa Kalb Smith. It was one of a set of books I received when I was in the Fifth Grade to help improve my reading skills.
Well, I could read quite well by the Second Grade. But, I wasn't reading fast enough for the nation-wide reading program in schools. I've carried that book with me wherever I have moved. And, I have moved many times over the years. My wife has counted seventeen moves over the forty-six years that we have been married!
I just wish I knew where Quiet Boy was right now. I'd like to take a picture of it to show you. That's the problem. I am somewhat in the same predicament as Old Mother Hubbard. I have so many books I don't know what to do to find them when I or another person wants them. What is the solution?
Locate and document the location in the Library Thing for use in TinyCat.
It's this last move five years ago that did me in. Before that, I always had the books group properly on the shelves and could find anything fairly quickly. This time, I was not incharge of packing and unpacking the books. I wasn't well enough. So they are all over the place.
One of the difficulties in getting their locations documented has been my increased difficulty with both handwriting and typing. I love to type and have been doing so most of my life. But, it is getting harder.
The Solution: Enter Baldrick's Cunning Plan
So, what is the answer? I will take pictures of each of the shelves. Then, I'll run them through AI to convert the pictures into text. Having coded the individual bookcases and their shelves, I will be able to add the locations to a downloaded copy of a Library Thing spreadsheet using Linux software. Simples. Well, there is some work involved.
To find out how much might be involved in just the picture-to-text portion of the project, I ran a small experiment. I picked three typical shelves in one of the bookcases and took a quick snap. Then, I ran it through an AI platform for which I have an assistive technology subscription.
The results were encouraging and you can view them here.
Was it perfect, no? Able to be improved, yes? There were a number of issues:
- The book spines were not uniformly presented on the shelf. Some were pushed further in than their companions. This affected both focus and lighting.
- Some of the books were slightly hidden in shadow due to books being piled on top of the row of books.
- The spines of older books were often difficult to discern as were some thinner books which had either very small type or no title on the spine at all.
There are a few steps to be taken to improve outcomes:
- Even out the books on the shelf and pull as far forward as possible. This is going to take
more work with the second rank of books behind the first as the books are double-shelved.
So, the first rank has to be pulled off, the second rank pulled forward, the picture taken, the second rank moved back and the first rank replaced in the proper position for photographing. All doable with some help...there are a lot of shelves around here. - Photograph the loungers (those books laying horizontally on top of the rows) spread out on a flat surface separately.
- Use a tripod stand for the camera rather than taking the photos freehand to improve stability and focus...stability is not something easily obtained when you have Parkinson's!
- Improve the lighting so that all of the spines are well lit.
- Manually tally the books with difficult to discern spines.
I will update progress in these regards soon!