What I Learned about Image Search this week
Last year roughly 20% of searches on Google were done by voice command.
That’s a lot of voice searches. People are becoming more and more comfortable talking to their devices. It’s safe to say that people are becoming just as comfortable with other forms of search. The search by picture function at most grocery store self-checkout kiosks is very effective. So what does Google have going in this arena?
It’s no surprise that Google has several fun tools related to image search. The first is being able to search by an actual image file. Full disclosure - I did not know Google had these capabilities, despite being an everyday user for many years.
Google has four ways for you to input an image file for it to search for related image files. Search by drag and drop, upload, copy and paste URL, and right click save as. It is a startlingly accurate function. A seemingly random image can result in a very specific result.
Another really great tool from Google is the Vision API.
I thought image analysis by itself was complicated. Vision also has image sentiment analysis. Not quite sure if Grandma looks happy in that family photo?
Google knows with 89% confidence.
Vision also allows you to pull color ratios and palettes, text analysis and even JSON data.
Considering the trend in voice search, it’s safe to say that image search will also catch on. I apply the model of the grocery store search by image to most retailers. Why search by SKU when one can simply search by what the product looks like? I also see Netflix as an image based movie reccomendation engine. Based on what each card looks like users curate their content. Expect to see more and more visual ways of searching and getting visual results.
