METHOD FOR THE AUTOMATED EXTRACTION OF A PLANOGRAM FROM IMAGES OF SHELVING
First Claim
1. Method for the automatic construction of a planogram, characterised in that it comprises:
- the reception of one or more shelving images (1),a step of assembling (2) said images, in the case where there are several,a step of automatic construction of the structure (3) constituting the planogram,a step of automatic recognition (4) of the products contained in the images,a step of positioning the products (5) according to the results of the previous automatic recognition of the products (4) and the results of the previous detection of the structure (3),the automatic recognition step (4) being itself characterised in that it comprises at a minimum the following two steps;
a) an initial recognition step (41), giving a set of possibilities, in the form of ordered pairs (hypothesis, probability of detection), in which a hypothesis is an ordered pair (position in the image, identified product). This step (41) consists of a category classification step (4171) and a step (4173) of global filtering of these results using probabilistic methods for integrating information specific to the products (I12),b) a second recognition step (42), using the results of (41) to make a new detection, consisting of a classification step (4271) based on matchings of points of interest, and a second filtering step (4273) at the planogram level, integrating global information on the context (I2) previously estimated on a base of planograms (I3),each of the classification and filtering steps (4171, 4173, 4271, 4273) being followed immediately by a step of selecting the best candidates, respectively (4172, 4174, 4272, 4274) corresponding to a thresholding of the detection probabilities.
3 Assignments
0 Petitions
Accused Products
Abstract
A method for automatically constructing a planogram from photographs of shelving, replacing laborious manual construction includes the following steps: a step (1) in which the images are received, a step (2) in which the images are assembled, a step (3) in which the structure is automatically constructed, a step (4) in which the products are automatically detected, and a step (5) in which the products are positioned in the structure. The product detection step (4) enhances traditional image recognition techniques, using artificial learning techniques to incorporate characteristics specific to the planograms. This product detection step (4) also includes at least two successive classification steps, namely: an initialization step (41) with detection of product categories; and a classification step (42) with the classification of the products themselves, each of these steps including a first image recognition step, followed by a statistical filtering step based on the characteristics specific to the planograms.
50 Citations
17 Claims
-
1. Method for the automatic construction of a planogram, characterised in that it comprises:
-
the reception of one or more shelving images (1), a step of assembling (2) said images, in the case where there are several, a step of automatic construction of the structure (3) constituting the planogram, a step of automatic recognition (4) of the products contained in the images, a step of positioning the products (5) according to the results of the previous automatic recognition of the products (4) and the results of the previous detection of the structure (3), the automatic recognition step (4) being itself characterised in that it comprises at a minimum the following two steps; a) an initial recognition step (41), giving a set of possibilities, in the form of ordered pairs (hypothesis, probability of detection), in which a hypothesis is an ordered pair (position in the image, identified product). This step (41) consists of a category classification step (4171) and a step (4173) of global filtering of these results using probabilistic methods for integrating information specific to the products (I12), b) a second recognition step (42), using the results of (41) to make a new detection, consisting of a classification step (4271) based on matchings of points of interest, and a second filtering step (4273) at the planogram level, integrating global information on the context (I2) previously estimated on a base of planograms (I3), each of the classification and filtering steps (4171, 4173, 4271, 4273) being followed immediately by a step of selecting the best candidates, respectively (4172, 4174, 4272, 4274) corresponding to a thresholding of the detection probabilities. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
Specification