Over the past few days I’ve been working on analyzing main causes of some app issues we’ve been having. With a combination of Splunk, Excel-foo, and some grunt-work, I dug through millions of log statements to find then analyze tens of thousands of error messages. I eventually came up with some pretty sweet looking charts to summarize my results.
Now I have tens of thousands of images I would like to analyze. How the heck can I do this? I have no idea. How can I evaluate and quantify if an image is blurry, too-low contrast, or off the edge of the image? Just by looking at some by hand, I can see how some would be rejected but how maybe we could change some parameters to accept more. How can I try making a change to our processing and run these thousands – and thousands of currently working images – back through to reevaluate the results?
Text is easy, but these images have me stumped.