Artificial intelligence-based system and method for grading collectible trading cards
View Patent ↗A system and method for digitally grading collectible trading cards on a predefined standard scale. The collectible trading cards are graded using their images. First, an image is converted to grayscale image. The grayscale image is subjected to a set of algorithms, such as edge detection algorithm, threshold inversion algorithm, wavelet transform algorithm, corner detection algorithm, color filtering algorithm, and an image sharpen algorithm to obtain respective image features as outputs. The output can be processed using a bag of visual words computer vision model to obtain quantitative data. The quantitative data can then be processed using a pre-trained machine learning model to obtain a grade for the collectible trading card.
1. A method for grading a collectible trading card using a combination of image recognition algorithms and artificial intelligence on a predefined scale, the method implemented on a system comprising a processor and a memory, the method comprises:
receiving, an image of the collectible trading card from a user device;
converting the image to a grayscale image;
applying an edge detection algorithm to the grayscale image to extract edge features;
applying a threshold inversion algorithm to the grayscale image to extract contrast/centering features;
applying a wavelet transform algorithm to the grayscale image to extract texture/surface features;
applying a corner detection algorithm to the grayscale image to extract corner information;
applying a color filtering algorithm to the grayscale image to extract stain detection features;
applying an image sharpen algorithm to the grayscale image to obtain an output and comparing the output with the image to obtain out-of-focus information;
processing the edge features, contrast/centering features, texture/surface features, corner information, stain detection features, and out-of-focus information using a bag-of-visual-words model to obtain quantitative data; subjecting the quantitative data to a pre-trained machine learning model to obtain a grade for the collectible trading card, wherein the grade is associated to the collectible trading card;
training a machine learning model using a set of pre-graded training images of collectible trading cards to obtain the pre-trained machine learning model; converting each training image of the pre-graded training images to a grayscale training image;
processing each grayscale training image using a plurality of predefined algorithms to obtain a plurality of outputs, the plurality of predefined algorithms comprises the edge detection algorithm, the threshold inversion algorithm, the wavelet transform algorithm, the corner detection algorithm, the color filtering algorithm, and the image sharpen algorithm;
subjecting the plurality of outputs to the bag-of-visual-words model to obtain respective quantitative data for each output;
generating a feature vector for each output using the respective quantitative data, wherein the feature vector is configured to grade the respective output based on the respective quantitative data.
2. The method according to claim 1 , wherein the method further comprises:
capturing the image of the collectible trading card by a camera coupled to the user device.
3. A system for grading a collectible trading card using a combination of image recognition algorithms and artificial intelligence on a predefined scale, the system comprising a processor and a memory, the system configured to implement a method, the method comprises:
receiving, an image of the collectible trading card from a user device;
converting the image to a grayscale image;
applying an edge detection algorithm to the grayscale image to extract edge features;
applying a threshold inversion algorithm to the grayscale image to extract contrast/centering features;
applying a wavelet transform algorithm to the grayscale image to extract texture/surface features;
applying a corner detection algorithm to the grayscale image to extract corner information;
applying a color filtering algorithm to the grayscale image to extract stain detection features;
applying an image sharpen algorithm to the grayscale image to obtain an output and comparing the output with the image to obtain out-of-focus information;
processing the edge features, contrast/centering features, texture/surface features, corner information, stain detection features, and out-of-focus information using a bag-of-visual-words model to obtain quantitative data;
subjecting the quantitative data to a pre-trained machine learning model to obtain a grade for the collectible trading card, wherein the grade is associated to the collectible trading card;
training a machine learning model using a set of pre-graded training images of collectible trading cards to obtain the pre-trained machine learning model;
converting each training image of the pre-graded training images to a grayscale training image;
processing each grayscale training image using a plurality of predefined algorithms to obtain a plurality of outputs, the plurality of predefined algorithms comprises the edge detection algorithm, the threshold inversion algorithm, the wavelet transform algorithm, the corner detection algorithm, the color filtering algorithm, and the image sharpen algorithm;
subjecting the plurality of outputs to the bag-of-visual-words model to obtain respective quantitative data for each output; and
generating a feature vector for each output using the respective quantitative data,
wherein the feature vector is configured to grade the respective output based on the respective quantitative data.
4. The system according to claim 3 , wherein the method further comprises:
capturing the image of the collectible trading card by a camera coupled to the user device.