Line 2: Line 2:
 
        
 
        
  
In the summer of 2009, I got the opportunity to work in the VIPER Lab at Purdue University under the direction of Professor Delp. The project I worked on is called "TADA" - a way of conducting dietary assessments on a mobile device. I performed the segmentation of the food images. An algorithm was written  which automatically processes images of a meal and segments it into the various components of the meal. Then on each segmented image, signal processing concepts like volume estimation is performed and then the results are sent to a server ,from where it can access a dietary database and the user can get information about the various nutrient information pertaining to their meal. I used Adobe photoshop to segment the images. I had to  keep track of the object boundaries while segmenting it. While processing an image ,its very essential to work with objects with finite boundaries,so that the threshold for recognizing the object is determined correctly and further, the computer can do a good job at automatically detecting the object. My segmented images were compared with the images which the algorithm segmented and  we used that analysis to see the quality of automatic segmentation done by the algorithm was close to the segmentation done by hand.
+
In the summer of 2009, I got the opportunity to work in the [[http://cobweb.ecn.purdue.edu/~ips/]]VIPER Lab at Purdue University under the direction of Professor Delp. The project I worked on is called "TADA" - a way of conducting dietary assessments on a mobile device. I performed the segmentation of the food images. An algorithm was written  which automatically processes images of a meal and segments it into the various components of the meal. Then on each segmented image, signal processing concepts like volume estimation is performed and then the results are sent to a server ,from where it can access a dietary database and the user can get information about the various nutrient information pertaining to their meal. I used Adobe photoshop to segment the images. I had to  keep track of the object boundaries while segmenting it. While processing an image ,its very essential to work with objects with finite boundaries,so that the threshold for recognizing the object is determined correctly and further, the computer can do a good job at automatically detecting the object. My segmented images were compared with the images which the algorithm segmented and  we used that analysis to see the quality of automatic segmentation done by the algorithm was close to the segmentation done by hand.

Revision as of 11:49, 16 April 2010

My Summer 2009 Research Project with Professor Edward delp


In the summer of 2009, I got the opportunity to work in the [[1]]VIPER Lab at Purdue University under the direction of Professor Delp. The project I worked on is called "TADA" - a way of conducting dietary assessments on a mobile device. I performed the segmentation of the food images. An algorithm was written which automatically processes images of a meal and segments it into the various components of the meal. Then on each segmented image, signal processing concepts like volume estimation is performed and then the results are sent to a server ,from where it can access a dietary database and the user can get information about the various nutrient information pertaining to their meal. I used Adobe photoshop to segment the images. I had to keep track of the object boundaries while segmenting it. While processing an image ,its very essential to work with objects with finite boundaries,so that the threshold for recognizing the object is determined correctly and further, the computer can do a good job at automatically detecting the object. My segmented images were compared with the images which the algorithm segmented and we used that analysis to see the quality of automatic segmentation done by the algorithm was close to the segmentation done by hand.

Alumni Liaison

Prof. Math. Ohio State and Associate Dean
Outstanding Alumnus Purdue Math 2008

Jeff McNeal