Find solution
How to match a drawn Chinese character to a database of stroke animations?

Is it possible to code a algorithm such that when a input image is given and the apparel which a person wears in the image is cropped and compared with a image in a database and come out with the exact match/closest match?

  • What is the higher end technology available in image processing to do this. If my image is a photo with some 3 friends in it and is it possible to do the same for such a image? And if my database image is a constant one and my input image is one with many variations in dimensions and other stuff is it possible to do,in case if its a NO  what is the maximum possibility?

  • Answer:

    I do not know too much about the papers or algorithms that do this. The one paper I do know is from Tamara Berg's group at Stony Brook, NY: http://tamaraberg.com/papers/parsingclothing.pdf They consider the task of parsing the clothing worn by people in a photograph and retrieving images from a database containing similar apparel. They claim to be able to handle variations in pose etc, so I think this paper is exactly what you are looking for. Their algorithm at a high level is simple: Segment out the image into regions (somewhat a solved problem), estimate the pose of the person (not solved, but usable algorithms exist), use the pose and the segmentation to get the various garments the person is wearing (eg. shirt, shorts etc) and retrieve similar garments from the dataset using color. So in principle this algorithm can do what you want (they also have code available). In practice, pose estimation is a hard problem, and if multiple people are huddling together it can get confused. The actual efficacy of the algorithm is something you will have to try out and check.

Bharath Hariharan at Quora Visit the source

Was this solution helpful to you?

Other answers

You might want to read this response by the Plink founder Your question is very similar, if not exact. There are plenty of publications and lots of research on this topic, search for Content Based Image Retrieval (CBIR). HTH

Shashi Kant

There are two questions I would deal with here. 1. Is it possible to code an algorithm such that you can get the apparel a person is wearing? Yes. There is an extensive research on cropping irregular, undefined shapes already done. They use the methods of active contouring and graph cut segmentation. In the former, usually an energy function is optimized to obtain the contour of the object. In graph cut segmentation, you divide your image into K clusters and perform the graph cut algorithm from there to segment your image. 2. Can you do it on multiple images, to obtain multiple segmentation, yes. I did not understand the concept about variation in dimensions. If you follow these papers and their references, you should be able to understand how the idea. ========================================================== http://www.tc.umn.edu/~baixx015/Xue_IJCV.pdf http://pdf.aminer.org/000/292/851/demonstration_of_segmentation_with_interactive_graph_cuts.pdf

Anonymous

Related Q & A:

Just Added Q & A:

Find solution

For every problem there is a solution! Proved by Solucija.

  • Got an issue and looking for advice?

  • Ask Solucija to search every corner of the Web for help.

  • Get workable solutions and helpful tips in a moment.

Just ask Solucija about an issue you face and immediately get a list of ready solutions, answers and tips from other Internet users. We always provide the most suitable and complete answer to your question at the top, along with a few good alternatives below.