A System And A Method For Efficient Image Recognition

Cognizant Technology Solutions India Pvt. Ltd.
Granted App No 201941028755 Patent 537602 Computer Science

Patent information

IP India · live record
Application number
201941028755
Filing date
17 July 2019
Status
Granted
Legal status
Inforce
Publication number
04/2021
Publication type
INA
Invention field
Computer Science
Patent number
537602
Grant date
10 May 2024
Next renewal due
17 July 2034

Abstract

A SYSTEM AND A METHOD FOR EFFICIENT IMAGE RECOGNITION A system and method for grouping of similar image classes for image recognition is provided. The invention comprises extracting one or more features from multiple classes of images for determining a correlation value between each of the multiple classes of images based on assessment of the extracted features of each one of the classes of images with respect to other classes of images in the multiple classes of images. Further, the one class of image is grouped with the other class of image in the multiple classes of images to form one or more groups of super-classes of similar class of images based on analysis of the determined correlation values with respect to a pre-determined threshold value. An input image is recognized based on the formed groups of super-classes followed by sub¬class classification of the images.

Applicants

1
India
Cognizant Technology Solutions India Pvt. Ltd. Techno Complex, No. 5/535, Old Mahabalipuram Road, Okkiyam Thoraipakkam, Chennai

Inventors

4
India
Kumar VishalSH-8/2A -5K, Behind U.P. College, Uchawalodge, Varanasi-221002
India
Ritesh MishraFlat- A3, Fortuna Serenity Apartment, 6th ‘D’ Cross, Kaggadasapura, Bengaluru – 560093
India
Arvind Channarayapatna Srinivasa279, 4th Cross, Banashankari 3rd Stage, 3rd Phase, Channammanakere Achkut, Bengaluru – 560085
India
Venugopal Gundimeda8-3-677/45, First Floor, SKD Nagar, Yellareddy Guda, Hyderabad – 500073

Claims

20
1 A method for grouping of similar image classes for image recognition, the method is implemented by a processor executing instructions stored in a memory, the method comprising: extracting one or more features from multiple classes of images for determining a correlation value between each of the multiple classes of images based on assessment of the extracted features of each one of the classes of images with respect to other classes of images in the multiple classes of images; grouping the one class of image with the other class of image in the multiple classes of images to form one or more groups of super-class of similar class of images based on analysis of the determined correlation values with respect to a pre-determined threshold value; and recognizing an input image based on the formed groups of super-classes followed by sub-class classification of the images.
2 The method as claimed in claim 1, wherein the features from the multiple image classes are extracted by executing a pre-determined parameter on each of the classes of images utilizing an image recognition base model.
3 The method as claimed in claim 1, wherein the correlation value represents an estimation of correlation of the one class of image with respect to the another class of image in the multiple classes of images.
4 The method as claimed in claim 1, wherein the correlation values are computed in a matrix form, wherein the matrix 32 comprises rows and columns equal to the number of image classes analyzed for determining the correlation values.
5 The method as claimed in claim 4, wherein the groups of super-classes with correlated classes are formed if the correlation value of the one class of image with respect to the other class of image in a row of the matrix is higher than the pre-determined threshold value.
6 The method as claimed in claim 1, wherein the one or more image classes in the super-class group belongs only to a particular super-class group without overlapping with another super-class group of comparatively lower correlation.
7 The method as claimed in claim 1, wherein the groups of super-classes of correlated image classes are labelled for generating one or more new image recognition models for recognition of inputted images as super-classes.
8 The method as claimed in claim 1, wherein the pre¬determined threshold value is decreased for increasing the grouping of the one class of image with another class of image in the multiple classes of images.
9 The method as claimed in claim 8, wherein based on the increased grouping multiple groups of super-classes of similar class of images are formed.
10 The method as claimed in claim 7, wherein a best model is selected from the one or more new models for image recognition as super-class.
11 The method as claimed in claim 1, wherein the super-class of images recognized are classified into one or more 33 sub-classes at least by one of: a sub-class image recognition model; and an image classification technique.
12 A system for grouping of similar image classes for image recognition, the system comprising: a memory storing program instructions; a processor configured to execute program instructions stored in the memory; and an image recognition engine in communication with the processor and configured to: extract one or more features from multiple classes of images for determining a correlation value between each of the multiple classes of images based on assessment of the extracted features of each one of the classes of images with respect to other classes of images in the multiple classes of images; group the one class of image with the other class of image in the multiple classes of images to form one or more groups of super-classes of similar class of images based on analysis of the determined correlation values with respect to a pre-determined threshold value; and recognize an input image based on the formed groups of super-classes followed by sub-class classification of the images.
13 The system as claimed in claim 12, wherein the image recognition engine comprises a model generation unit in communication with the processor, the model generation unit is configured to generate a base model for image recognition, wherein the base model is utilized for feature extraction from the multiple classes of images. 34
14 The system as claimed in claim 12, wherein the image recognition engine comprises a correlation unit in communication with the processor, the correlation unit is configured to compute the correlation values in a matrix form, wherein the matrix comprises rows and columns equal to the number of image classes analyzed for determining the correlation values.
15 The system as claimed in claim 12, wherein the image recognition engine comprises a threshold generation unit in communication with the processor, the threshold generation unit is configured to analyze the correlation values present in each row of the matrix with a pre¬determined threshold value.
16 The system as claimed in claim 12, wherein the image recognition engine comprises a class grouping unit in communication with the processor, the class grouping unit is configured to group the one class of image with the other class of image to form a super-class of images if the correlation value of the one class of image with respect to the other class of image in a row of the matrix is higher than the pre-determined threshold value.
17 The system as claimed in claim 12, wherein the image recognition engine comprises a new model generation unit in communication with the processor, the new model generation unit is configured to generate a new image recognition model based on the labelled super-classes of images for image recognition of inputted images.
18 The system as claimed in claim 12, wherein the image recognition engine comprises a model comparison and selection unit in communication with the processor, the 35 model comparison and selection unit is configured to communicate with the threshold generation unit for decreasing the pre-determined threshold value for increasing the grouping of the one class of image with the other class of image in the multiple classes of images.
19 The system as claimed in claim 19, wherein the model comparison and selection unit is configured to communicate with the model generation unit and the new model generation unit for comparing the generated new image recognition model with respect to the base model for selecting a best image recognition model from the generated one or more new image recognition models.
20 The system as claimed in claim 12, wherein the image recognition engine comprises an image classification unit in communication with the processor, the image classification unit is configured to classify the super-classes of images into one or more sub-classes at least by one of: a sub-class image recognition model; and an image classification technique.

Specification

Show full textCollapse

We claim: 1. A method for grouping of similar image classes for image recognition, the method is implemented by a processor executing instructions stored in a memory, the method comprising: extracting one or more features from multiple classes of images for determining a correlation value between each of the multiple classes of images based on assessment of the extracted features of each one of the classes of images with respect to other classes of images in the multiple classes of images; grouping the one class of image with the other class of image in the multiple classes of images to form one or more groups of super-class of similar class of images based on analysis of the determined correlation values with respect to a pre-determined threshold value; and recognizing an input image based on the formed groups of super-classes followed by sub-class classification of the images. 2. The method as claimed in claim 1, wherein the features from the multiple image classes are extracted by executing a pre-determined parameter on each of the classes of images utilizing an image recognition base model. 3. The method as claimed in claim 1, wherein the correlation value represents an estimation of correlation of the one class of image with respect to the another class of image in the multiple classes of images. 4. The method as claimed in claim 1, wherein the correlation values are computed in a matrix form, wherein the matrix 32

comprises rows and columns equal to the number of image classes analyzed for determining the correlation values. 5. The method as claimed in claim 4, wherein the groups of super-classes with correlated classes are formed if the correlation value of the one class of image with respect to the other class of image in a row of the matrix is higher than the pre-determined threshold value. 6. The method as claimed in claim 1, wherein the one or more image classes in the super-class group belongs only to a particular super-class group without overlapping with another super-class group of comparatively lower correlation. 7. The method as claimed in claim 1, wherein the groups of super-classes of correlated image classes are labelled for generating one or more new image recognition models for recognition of inputted images as super-classes. 8. The method as claimed in claim 1, wherein the pre¬determined threshold value is decreased for increasing the grouping of the one class of image with another class of image in the multiple classes of images. 9. The method as claimed in claim 8, wherein based on the increased grouping multiple groups of super-classes of similar class of images are formed.

10. The method as claimed in claim 7, wherein a best model is selected from the one or more new models for image recognition as super-class. 11. The method as claimed in claim 1, wherein the super-class of images recognized are classified into one or more 33

sub-classes at least by one of: a sub-class image recognition model; and an image classification technique. 12. A system for grouping of similar image classes for image recognition, the system comprising: a memory storing program instructions; a processor configured to execute program instructions stored in the memory; and an image recognition engine in communication with the processor and configured to: extract one or more features from multiple classes of images for determining a correlation value between each of the multiple classes of images based on assessment of the extracted features of each one of the classes of images with respect to other classes of images in the multiple classes of images; group the one class of image with the other class of image in the multiple classes of images to form one or more groups of super-classes of similar class of images based on analysis of the determined correlation values with respect to a pre-determined threshold value; and recognize an input image based on the formed groups of super-classes followed by sub-class classification of the images. 13. The system as claimed in claim 12, wherein the image recognition engine comprises a model generation unit in communication with the processor, the model generation unit is configured to generate a base model for image recognition, wherein the base model is utilized for feature extraction from the multiple classes of images. 34

14. The system as claimed in claim 12, wherein the image recognition engine comprises a correlation unit in communication with the processor, the correlation unit is configured to compute the correlation values in a matrix form, wherein the matrix comprises rows and columns equal to the number of image classes analyzed for determining the correlation values. 15. The system as claimed in claim 12, wherein the image recognition engine comprises a threshold generation unit in communication with the processor, the threshold generation unit is configured to analyze the correlation values present in each row of the matrix with a pre¬determined threshold value. 16. The system as claimed in claim 12, wherein the image recognition engine comprises a class grouping unit in communication with the processor, the class grouping unit is configured to group the one class of image with the other class of image to form a super-class of images if the correlation value of the one class of image with respect to the other class of image in a row of the matrix is higher than the pre-determined threshold value. 17. The system as claimed in claim 12, wherein the image recognition engine comprises a new model generation unit in communication with the processor, the new model generation unit is configured to generate a new image recognition model based on the labelled super-classes of images for image recognition of inputted images. 18. The system as claimed in claim 12, wherein the image recognition engine comprises a model comparison and selection unit in communication with the processor, the 35

model comparison and selection unit is configured to communicate with the threshold generation unit for decreasing the pre-determined threshold value for increasing the grouping of the one class of image with the other class of image in the multiple classes of images. 19. The system as claimed in claim 19, wherein the model comparison and selection unit is configured to communicate with the model generation unit and the new model generation unit for comparing the generated new image recognition model with respect to the base model for selecting a best image recognition model from the generated one or more new image recognition models. 20. The system as claimed in claim 12, wherein the image recognition engine comprises an image classification unit in communication with the processor, the image classification unit is configured to classify the super-classes of images into one or more sub-classes at least by one of: a sub-class image recognition model; and an image classification technique.

Documents

Sign in to open
Application documents
1 201941028755-STATEMENT OF UNDERTAKING (FORM 3) [17-07-2019(online)].pdf 17 Jul 2019
2 201941028755-PROOF OF RIGHT [17-07-2019(online)].pdf 17 Jul 2019
3 201941028755-POWER OF AUTHORITY [17-07-2019(online)].pdf 17 Jul 2019
4 201941028755-FORM 1 [17-07-2019(online)].pdf 17 Jul 2019
5 201941028755-DRAWINGS [17-07-2019(online)].pdf 17 Jul 2019
6 201941028755-COMPLETE SPECIFICATION [17-07-2019(online)].pdf 17 Jul 2019
7 Correspondence by Agent_Form-1 And POA_22-07-2019.pdf 22 Jul 2019
8 201941028755-FORM 18 [23-07-2019(online)].pdf 23 Jul 2019
9 201941028755-Request Letter-Correspondence [24-07-2019(online)].pdf 24 Jul 2019
10 201941028755-Form 1 (Submitted on date of filing) [24-07-2019(online)].pdf 24 Jul 2019
11 201941028755-FORM 3 [09-10-2019(online)].pdf 09 Oct 2019
12 201941028755-FER.pdf 17 Oct 2021
13 201941028755-FORM 3 [05-01-2022(online)].pdf 05 Jan 2022
14 201941028755-FER_SER_REPLY [05-01-2022(online)].pdf 05 Jan 2022
15 201941028755-COMPLETE SPECIFICATION [05-01-2022(online)].pdf 05 Jan 2022
16 201941028755-CLAIMS [05-01-2022(online)].pdf 05 Jan 2022
17 201941028755-ABSTRACT [05-01-2022(online)].pdf 05 Jan 2022
18 201941028755-PatentCertificate10-05-2024.pdf 10 May 2024
19 201941028755-IntimationOfGrant10-05-2024.pdf 10 May 2024
Search strategy
1SearchStrategy201941028755E_08-07-2021.pdf

ERegister / renewals

Inforce
3rd Renewed 20 May 2024 Period 17 Jul 2021 – 17 Jul 2022 · CBR 33430 · Certificate 98920 ₹4,000
4th Renewed 20 May 2024 Period 17 Jul 2022 – 17 Jul 2023 · CBR 33430 · Certificate 98921 ₹4,000
5th Renewed 20 May 2024 Period 17 Jul 2023 – 17 Jul 2024 · CBR 33430 · Certificate 98922 ₹4,000
6th Renewed 20 May 2024 Period 17 Jul 2024 – 17 Jul 2025 · CBR 33430 · Certificate 98923 ₹4,000
7th Renewed 20 May 2024 Period 17 Jul 2025 – 17 Jul 2026 · CBR 33430 · Certificate 98924 ₹10,800
8th Renewed 20 May 2024 Period 17 Jul 2026 – 17 Jul 2027 · CBR 33430 · Certificate 98925 ₹10,800
9th Renewed 20 May 2024 Period 17 Jul 2027 – 17 Jul 2028 · CBR 33430 · Certificate 98926 ₹10,800
10th Renewed 20 May 2024 Period 17 Jul 2028 – 17 Jul 2029 · CBR 33430 · Certificate 98927 ₹10,800
11th Renewed 20 May 2024 Period 17 Jul 2029 – 17 Jul 2030 · CBR 33430 · Certificate 98928 ₹21,600
12th Renewed 20 May 2024 Period 17 Jul 2030 – 17 Jul 2031 · CBR 33430 · Certificate 98929 ₹21,600
13th Renewed 20 May 2024 Period 17 Jul 2031 – 17 Jul 2032 · CBR 33430 · Certificate 98930 ₹21,600
14th Renewed 20 May 2024 Period 17 Jul 2032 – 17 Jul 2033 · CBR 33430 · Certificate 98931 ₹21,600
15th Renewed 20 May 2024 Period 17 Jul 2033 – 17 Jul 2034 · CBR 33430 · Certificate 98932 ₹21,600