Artificial Intelligence No longer Science Fiction.
One of the most transformative developments in artificial intelligence is in the sub field of computer vision. While computer vision has been available commercially for several decades, it is experiencing a renaissance today as a result of new algorithms and the availability of low-cost, cloud-based super-computing. Today, identifying specific characteristics of a photo using computer vision can be realized with such remarkable speed and precision that it is transforming how many enterprises monetize their visual content. Only a few years ago, this was still the province of academic curiosity.
Ditto Labs Early Innovator and Consistent Leader.
Since our first product launch, Ditto Labs has been processing vast streams of social media photos on behalf of marketing professionals. Our clients are interested in identifying specific brands, facial expressions, objects, and scenes within social posts. More recently, Ditto’s client base has expanded to include companies with their own massive databases of proprietary photos. These clients can not unlock the full value of their visual assets without humans in the loop to filter, prioritize and label their attributes. On-line merchants, or so called e-tailers, experience higher click through rates when the photo illustrating a product is perceived to be authentic by their audience. And so merchants include user generated content to sell all sorts of goods, hotels, restaurants, etc. These clients have the operational challenge to keep their content fresh and of high quality without maintaining a human reviewer in the loop.
image
Beneficiaries in Travel and Hospitality.
For example, today travel and restaurant sites encourage hotel guests and restaurant goers to share their experiences by posting photos and reviews. Hotel chains like Marriott, Starwood, Hilton, Wyndham and Hyatt encourage the same behavior–they see the photos-as-feedback as one part customer loyalty and one part customer service. The evergreen flow of new photos burdens them with the untenable requirement to review each. This is a costly, labor-intensive, and time-consuming process that invites significant brand risk. The more successful the website, the higher the volume of photos and therefore the more challenging it is to keep up with human reviewers.
Computer Vision Offers an Automated Solution.
And Ditto Labs is a leader in this burgeoning field as we enable our clients to quickly develop and deploy a set of classifiers to sort their millions of photos by the attributes which are most important to their business and most aligned with editorial guidelines. Perhaps most compelling is Ditto’s unique ability to quickly train on our clients’ proprietary data and deliver bespoke classifiers that tag, filter, and organize photos in ways that are most valuable to them.
image
Ditto uses a proprietary active learning loop to improve classifier performance over time.
Custom Solutions Learn From Your Data.
Ditto trains classifiers to recognize specific scenes like hotel lobby, pool patio or bathroom shower, or specific objects like couch, canopy bed, salad or sushi. Perhaps most compelling to our clients is our ability to train classifiers to detect if a photo contains a conceptual description like “messy,” “modern” or “partially eaten.” With Ditto’s tool kit, editing out photos that are out of focus, contain nudity or are of “low visual quality” is automated so those which score as the most beautiful, crisp can be featured and interest in a particular property can include an appropriately diverse representation of authentic photos from customers.
image
Example of precision-recall curve for all iterations on the current test set.
That’s Money in Your Pocket.
Our clients experience higher click-through rates with more compelling, relevant visual content and they eliminate the cost of the manual reviewers. And, customers have a more satisfying experience and so their brand is fortified as well.
Come see for yourself!
The post Ditto for Travel and Hospitality appeared first on Ditto Labs, Inc..