Google Images

In 2000, Google Search results were limited to simple pages of text with links. Google's developers worked on developing this further, and they realized that an image search tool was required to answer "the most popular search query" they had seen to date: the green Versace dress of Jennifer Lopez worn in February 2000.[5] Google paired a recently hired engineer Huican Zhu with product manager Susan Wojcicki (now the current CEO of YouTube) to build the feature, and they launched Google Image Search in July 2001.[6] That year, 250 million images were indexed in Image Search. This grew to 1 billion images by 2005 and over 10 billion images by 2010.[7]

In January 2007, Google updated the interface for the image search, where information about an image, such as resolution and URL, was hidden until the user moved the mouseover its thumbnail. This was discontinued after a few weeks.[8]

On October 27, 2009, Google Images added a feature to its image search that can be used to find similar images.[9]

On July 20, 2010, Google made another update to the interface of Google Images, which hid image details until mouseover.[10]

In May 2011, Google introduced a sort by subject feature for a visual category scheme overview of a search query.[11]

In June 2011, Google Images added a "Search by Image" feature which allowed for reverse image searches directly in the image search-bar without third-party add-ons. This feature allows users to search for an image by dragging and dropping one onto the search bar, uploading one, or copy-pasting a URL that points to an image into the search bar.[12]

On December 11, 2012, Google Images' search engine algorithm was changed once again, in the hopes of preventing pornographic images from appearing when non-pornographic search terms were used.[13][14] According to Google, pornographic images would still appear as long as the term searched for was specifically pornographic; otherwise, they would not appear. While Google stated explicitly that they were "not censoring any adult content", it was immediately noted that even when entering terms such as or "Breast," no explicit results were shown.[15][16][17] The only alternative option was to turn on an even stricter filter which would refuse to search for the aforementioned terms whatsoever.[17] Users could also no longer exclude keywords from their searches.[18]

On February 15, 2018, the interface was modified to meet the terms of a settlement and licensing partnership with Getty Images. The "View image" button (a deep link to the image itself on its source server) was removed from image thumbnails. This change is intended to discourage users from directly viewing the full-sized image (although doing so using a browser's context menu on the embedded thumbnail is not frustrated), and encourage them to view the image in its appropriate context (which may also include attribution and copyright information) on its respective web page. The "Search by image" button has also been downplayed, as reverse image search can be used to find higher-resolution copies of copyrighted images. Google also agreed to make the copyright disclaimer within the interface more prominent.[19]

On August 6, 2019, the ability to filter images by their image resolutions was removed, as well as "larger than", "face", and "full color" filters.[20]

The relevancy of search results has been examined. Most recently (October 2022), it was shown that 93.1% images of 390 anatomical structures were relevant to the search term.[21]

The general steps that Search by Image takes to get from a submitted image to returned search results are as follows:[23]

1. Analyze image: The submitted image is analyzed to find identifiers such as colors, points, lines, and textures.
2. Generate query: These distinct features of the image are used to generate a search query.
3. Match image: The query is matched against the images in Google's back end.
4. Return results: Google's search and match algorithms return matching and visually similar images as results to the user.