Pirelli Tyre S.p.a.

Datalogic IP Tech S.r.l.

Stmicroelectronics S.r.l.

Alma Mater Studiorum-Universitá Di Bologna

Luigi Di Stefano

Vincenzo Boffa

Valeriano Ballardini

Fabio Regoli

Alessandro Ghidotti Piovan

Alessandro Franchi

Federico Tombari

Danilo Pietro Pau

Giuseppe Casadio Tozzi

Alioscia Petrelli

Daniele De Gregorio

It is described an image labeling system () comprising: a support () for an object () to be labeled; a digital camera () configured to capture a plurality of images of a scene including said object (); a process and control apparatus () configured to receive said images and generate corresponding labeling data (-, L-L) associated to said object (); a digital display () associated with said support () and connected to the process and control apparatus () to selectively display additional images (-) selected from the group comprising: first images (-) in the form of backgrounds for the plurality of images and introducing a degree of variability in the scene; second images () indicating position and/or orientation according to which place said object () by a user on the support (); third images () to be captured by the digital camera () and provided to the process and control apparatus () to evaluate a position of the digital camera () with respect the digital display (); fourth images to be captured by the digital camera () and provided to the process and control apparatus () to evaluate at least one of the following data of the object (): position, orientation, 3D shape.

Creating training data variability in machine learning for object labelling from images

Method and related apparatus for detecting defects on a surface of a tyre, comprising: providing the tyre (); acquiring a digital image comprising a structure comprising sections representative of linear elements of a pattern in a surface portion and representative of possible elongated defects, said sections of the structure having a respective orientation; providing a model of the pattern in the surface portion, wherein each pixel is associated with a first index representative of whether the pixel belongs or not to a pattern section and a second index representative of an at least local orientation of the pattern section passing through said pixel; calculating for each pixel of the structure a third index representative of the orientation of the structure section passing through said pixel; and establishing, for each pixel of the structure having a corresponding pixel in the pattern model belonging to the pattern, whether said pixel of the structure belongs to a proposed defect on the basis of the comparison between the third index and the second index associated with the corresponding pixel in the pattern model.

Method and apparatus for detecting defects on the surface of tyres

Method and related apparatus for analysing a surface of a tyre, includes: acquiring one or more digital images of a surface portion having a pattern that includes a scheme that is repeated substantially equal thereto in a plurality of positions; identifying a plurality of first regions of said one or more digital images, each corresponding to a respective scheme sub-portion; identifying a respective plurality of regions homologous to each first region, wherein the respective scheme sub-portion is substantially identical to the respective scheme sub-portion of each first region; calculating a model of the respective scheme sub-portion, where each pixel is associated with a mean value of the values associated with the pixels of each first region and of the respective homologous regions having the same relative coordinates of said each pixel; and obtaining a pattern model using the calculated scheme sub-portion models.

Method and apparatus for analyzing a surface of a tyre

Local descriptors are extracted from digital image information and digital depth information related to digital images. The local descriptors convey appearance description information and shape description information related to the digital images. Global representations of the one or more digital images are generated based on the extracted local descriptors, and are hashed. Visual search queries are generated based on the hashed global representations. The visual search queries include fused appearance description information and shape description information conveyed in the local descriptors. The fusing may occur before the global representations are generated, before the hashing or after the hashing.

Method for visual search, corresponding system, apparatus and computer program product

Method and related apparatus for detecting defects on a surface of a tire, include: providing the tire; acquiring a digital image including a structure including sections representative of linear elements of a pattern in a surface portion and representative of possible elongated defects, the sections of the structure having a respective orientation; providing a model of the pattern in the surface portion, wherein each pixel is associated with a first index representative of whether the pixel belongs or not to a pattern section and a second index representative of an at least local orientation of the pattern section passing through the pixel; calculating for each pixel of the structure a third index representative of the orientation of the structure section passing through the pixel; and establishing, for each pixel of the structure having a corresponding pixel in the pattern model belonging to the pattern, whether the pixel of the structure belongs to a proposed defect on the basis of the comparison between the third index and the second index associated with the corresponding pixel in the pattern model.

A method for detecting defects on tires in a tire production process includes providing a tire; acquiring a three-dimensional image of a surface portion of the tire; generating, as a function of the acquired image, a plurality of values indicating the measure of a height profile of the surface portion of the tire; calculating, as a function of the plurality of values of the measure of the height profile and according to a consensus interpolation, a plurality of values indicating an estimation of a reference height profile of the surface portion of the tire; calculating, as a function of the plurality of values of the measure of the height profile and of the estimation of the reference height profile, a height profile of possible defects in the surface portion of the tire; and comparing values of the height profile of possible defects with respect to a threshold value in order to detect possible defects in the surface portion of the tire.

Method and apparatus for detecting defects on tyres in a tyre production process

During a description technique, a local descriptor for an object may be generated by computing a 2-dimensional histogram of pairs of angles between pairs of line segments that are aligned with edge pixels associated with the object. The pairs of line segments belong to a subset of k neighboring or proximate line segments. Moreover, this 2D histogram may represent the relative displacement and the relative orientations of the pairs of line segments in the subset as weights in bins or cells defined by angular quantization values, and the 2D histogram may exclude lengths of the line segments. Subsequently, the generated 2D histogram may be compared to predefined sets of descriptors in a model library that are associated with a set of objects, and the object may be identified as one of the set of objects based on a group of match scores determined in the comparisons.

Describing objects using edge-pixel-feature descriptors

During a description technique (), a local descriptor for an object () is generated () by computing a 2-dimensional histogram () of pairs of angles () between pairs of line segments () that are aligned with edge pixels associated with the object (). The pairs of line segments () belong to a subset of k neighboring or proximate line segments (). Moreover, this 2D histogram () may represent the relative displacement and the relative orientations of the pairs of line segments () in the subset as weights in bins or cells defined by angular quantization values, and the 2D histogram (may exclude lengths of the line segments. Subsequently, the generated 2D histogram () may be compared () to predefined sets of descriptors in a model library that are associated with a set of objects, and the object may be identified () as one of the set of objects based on a group of match scores determined in the comparisons.

Growing community of inventors

Bologna, Italy

Luigi Di Stefano