System & method to identify the individual products in a complex image in the e-commerce system & to output the product search results based on this identification.
Publication Date: 2016-Sep-08
The IP.com Prior Art Database
Disclosed is a system & method which gives an option for the user to upload a complex image consisting of many objects in the complex image or to capture a part of the complex image & to upload it & this system placed on the server side of e-commerce website internally identifies each object (product) name present in the uploaded complex image & then gives the product search result on an e-commerce website.
Page 01 of 3
System & method to identify the individual products in a complex image in the e-commerce system & to output the product search results based on this identification .
In today's world, shopping on e-commerce websites has become popular because of ease of shopping at home & the wide variety of product choices available on these websites.
But sometimes we find it difficult to find a product because we won't be knowing the name of the produce itself. This happens when we see a new product on some person or at some places or when we are seeing different geographical region's product & different cultured people's product. In another cases, we will be knowing the name of the product in our native language but still we may not be knowing the name of the product in the local language or the language used by the e-commerce websites.
This is explained with the following usecase.
Consider a foreigner who is in India & who is unaware of Indian culture & names. But he has seen few dresses in a wedding function.
Assume that he has liked the "Sherwani" dress but he does not know the name of it to search on the e-commerce websites.
But he has image of the person who was wearing it.
So in this case, the foreigner will find it difficult to know the name of the product (which in this case is "Sherwani").
He has to ask few people to find the name of the product by describing them about the dress or have to show the image or he has to search on the internet to find it.
So this is the problem in the existing solutions that there are no such e-commerce sites which provides image based search to identify products in a complex image.
The above is a simple example but there could be more complex scenarios in which we couldn't identify objects worn by person in the image because those objects could be new to the market or we are unaware of their ethnic wears.
Key feature is to have a system in place which accepts a complex image as it's input & it breaks down this complex image into it's individual parts containing each object using it's own internally programmed mechanism.
After this, it will start searching each divided image associated with a particular object to find the name of the product.
This system will have it's own internal image repository which will have images associated with most of the objects available in the market. So, it will use this image repository to find the match of the image & thus to find the name of the product.
This system will also have another option to the user in which the user can capture a particular part of the complex image to be given as the input image for finding the product name.
Page 02 of 3
Figure 1 shows how the e-commerce website's front-end looks like which has implemented this system.
Page 03 of 3
Figure 2 shows the architectural block diagram o...