Veggie Vision by IBM

21
CSE 803 Fall 2015 1 Veggie Vision by IBM Ideas about a practical system to make more efficient the selling an inventory of produce in a grocery sto

description

Ideas about a practical system to make more efficient the selling and inventory of produce in a grocery store. Veggie Vision by IBM. Problem is recognizing produce. properly charge customer do inventory save customer and checker time. 15+ years of R&D now. - PowerPoint PPT Presentation

Transcript of Veggie Vision by IBM

Page 1: Veggie Vision by IBM

CSE 803 Fall 2015 1

Veggie Vision by IBM

Ideas about a practical system to make more efficient the selling and

inventory of produce in a grocery store.

Page 2: Veggie Vision by IBM

CSE 803 Fall 2015 2

Problem is recognizing produce

•  properly charge customer •  do inventory •  save customer and checker time

Page 3: Veggie Vision by IBM

CSE 803 Fall 2015 3

15+ years of R&D now This information was shared by IBM researchers. Since that time, the system has been tested in small markets and has been modified according to that experience.

Page 4: Veggie Vision by IBM

CSE 803 Fall 2015 4

Up to 400 produce types

Page 5: Veggie Vision by IBM

CSE 803 Fall 2015 5

Practical problems of application environment

Page 6: Veggie Vision by IBM

CSE 803 Fall 2015 6

Engineering the solution

Page 7: Veggie Vision by IBM

CSE 803 Fall 2015 7

System to operate inside the usual checkout station

•  together with bar code scanner •  together with scale •  together with accounting •  together with inventory •  together with employee •  within typical store environment * figure shows system asking for help from the cashier in making final decision on touch screen

Page 8: Veggie Vision by IBM

CSE 803 Fall 2015 8

Modifying the scale

Page 9: Veggie Vision by IBM

CSE 803 Fall 2015 9

Need careful lighting engineering

Page 10: Veggie Vision by IBM

CSE 803 Fall 2015 10

Need to segment product from background, even through plastic

Page 11: Veggie Vision by IBM

CSE 803 Fall 2015 11

Previously published thresholding decision

Page 12: Veggie Vision by IBM

CSE 803 Fall 2015 12

Quality segmented image obtained

Page 13: Veggie Vision by IBM

CSE 803 Fall 2015 13

Design of pattern recognition paradigm (from 1997)

FEATURES are: color, texture, shape, and size all represented uniformly by HISTOGRAMS Histograms capture statistical properties of regions – any number of regions.

Page 14: Veggie Vision by IBM

CSE 803 Fall 2015 14

Matching procedure n  Sample product represented by concatenated

histograms: about 400 D n  350 produce items x 10 samples = 3500

feature vectors of 400D each n  Have about 2 seconds to compare an

unknown sample to 3500 stored samples (3500 dot products)

n  Analyze the k nearest: if closest 2 are from one class, recognize that class (sure)

Page 15: Veggie Vision by IBM

CSE 803 Fall 2015 15

HSI for pixel color: 6 bits for hue, 5 for saturation and intensity

For each pixel quantify H HIST[H]++ same for S&I

Page 16: Veggie Vision by IBM

CSE 803 Fall 2015 16

Histograms of 2 limes versus 3 lemons

Distribution or population concept adds robustness: •  to size of objects •  to number of objects •  to small variations of color (texture, shape, size)

Page 17: Veggie Vision by IBM

CSE 803 Fall 2015 17

Texture: histogram results of LOG filter[s] on produce pixels

Leafy produce A

Leafy produce B

Page 18: Veggie Vision by IBM

CSE 803 Fall 2015 18

Shape: histogram of curvature of boundary of produce

Page 19: Veggie Vision by IBM

CSE 803 Fall 2015 19

Banana versus lemon or cucumber versus lime

Large range of curvatures indicates complex object

Small range of curvatures indicates roundish object

Page 20: Veggie Vision by IBM

CSE 803 Fall 2015 22

Learning and adaptation n  System “easy” to train: show it produce

samples and tell it the labels. n  During service: age out oldest sample;

replace last used sample with newly identified one.

n  When multiple labeled samples match the unknown, system asks cashier to select from the possible choices.

Page 21: Veggie Vision by IBM

CSE 803 Fall 2015 23

Extension of Veggie Vision n  http://www.internetnews.com/xSP/article.php/

3642386 n  System uses almost all color features n  Installed in few places: many stores have self-

checkout, putting work on customer. n  IBM has a “shopping research” unit

http://www.usatoday.com/tech/news/techinnovations/2003-09-26-future-grocery-shop_x.htm

n  Customers will tolerate a higher human error than a machine error