Browse Prior Art Database

A System to retrieve photos without face recognition

IP.com Disclosure Number: IPCOM000244213D
Publication Date: 2015-Nov-24
Document File: 4 page(s) / 166K

Publishing Venue

The IP.com Prior Art Database

Abstract

Today, due to the wide usage of digital camera and large storage, people have huge amounts of photos on personal devices or cloud storage. Photo retrieval become a challenge. Existing photo retrieval products mainly focus on face recognition. However, sometimes a user may be interested in photos containing a specific person who is back to the camera. This article provide an approach to retrieve photos by recognizing special characteristic on target person, e.g. color of clothes, with or without hats, and trademarks on the clothes etc. Moreover, our approach can refine the results based on users’ feedback.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 4

A System to retrieve photos without face recognition

Today, due to the wide usage of digital camera and large storage, people have huge amounts of photos on personal device or cloud storage. Photo retrieval become a challenge. One example is, my son's kindergarten always take thousands of photos during school activities, and publish them on Cloud without category. It is difficult for the parent to find photo of their baby.

Even though some Cloud platforms provide Face Recognition function, the recognition rate is low because it only recognize clear front face but missing face side, portrait over long distance and so on. As the example, a mother like me not only interested in image which show my kid's face, but also want to see his state, behaviors, interaction with other kids. In some photos he maybe back to the camera, but I still want to see it. Current solution fail to satisfy such requirement. I have to review all the photos one by one to find what I want.

Our approach is to retrieval photos based on people's Clothing feature, such as type(T-shirt, dress) and color of clothing/shoes/hat/bag, special mark on clothing, etc. It will help people to retrieve the photos including a person with special feature, even if he is far from the focus, and even he only have a foot or a back view in the photo.

Figure 1 is a use case of our approach.

Step1: User choose photo resources.

Step2: System analyzes photo to retrieve persons and their clothing feature, store in database.

Step3: User chooses types and color of clothing to match. In this case u...