Deep-learning-based system and process for image recognition
Disclosed are methods and systems for using artificial intelligence (AI) for image recognition by using predefined coordinates to extract a portion of a received image, the extracted portion comprising a word to be identified having at least a first letter and a second letter; executing an image recognition protocol to identify the first letter; when the server is unable to identify the second letter, the server executes an AI model having a nodal data structure to identify the second letter based upon the identified first letter, the nodal data structure comprising a set of nodes where each node represents a letter, each node connected to at least one other node, wherein connection of a first node to a second node corresponds to a probability that a letter corresponding to the second node is used in a word subsequent to a letter corresponding to the first node.
1. A method comprising:
executing, by a processor, an image recognition protocol purported to identify a letter within a word;
in the event the image recognition protocol fails to identify the letter within the word, executing, by the processor, a neural network comprising a nodal data structure that identifies the letter within the word based upon a preceding letter within the word depicted within an image,
the nodal data structure comprising a set of nodes where each node represents a letter, each node within the set of nodes connected to at least one other node within the set of nodes, wherein connection of a first node to a second node corresponds to a probability that a letter corresponding to the second node is used subsequent to a preceding letter corresponding to the first node, wherein the processor queries an n-gram database to retrieve the probability of the letter being used after the preceding letter;
transmitting, by the processor, an identification of the letter to a second processor.
2. The method of claim 1 , wherein the processor extracts the word from the image using an optical character recognition protocol.
3. The method of claim 1 , wherein the word is depicted within predefined coordinates of the image.
4. The method of claim 1 , wherein the image is a check image.
5. The method of claim 1 , further comprising:
removing, by the processor, visual noise from the image.
6. The method of claim 1 , further comprising:
de-slanting, by the processor, at least a portion of the image.
7. The method of claim 5 , wherein the visual noise is a line that is not part of the word.
8. The method of claim 1 , further comprising:
training, by the processor, the neural network when the second processor indicates that the letter is correct or incorrect.
9. A system comprising:
a server comprising a processor and a non-transitory computer-readable medium containing instructions that when executed by the processor causes the processor to perform operations comprising:
executing an image recognition protocol purported to identify a letter within a word;
in the event the image recognition protocol fails to identify the letter within the word, executing a neural network comprising a nodal data structure to identify the letter within the word based upon a preceding letter within the word depicted within an image,
the nodal data structure comprising a set of nodes where each node represents a letter, each node within the set of nodes connected to at least one other node within the set of nodes, wherein connection of a first node to a second node corresponds to a probability that a letter corresponding to the second node is used subsequent to a preceding letter corresponding to the first node, wherein the processor queries an n-gram database to retrieve the probability of the letter being used after the preceding letter;
transmitting an identification of the letter to a second processor.
10. The system of claim 9 , wherein the processor extracts the word from the image using an optical character recognition protocol.
11. The system of claim 9 , wherein the word is depicted within predefined coordinates of the image.
12. The system of claim 9 , wherein the image is a check image.
13. The system of claim 9 , wherein the instructions further cause the processor to remove visual noise from the image.
14. The system of claim 9 , wherein the instructions further cause the processor to de-slant at least a portion of the image.
15. The system of claim 13 , wherein the visual noise is a line that is not part of the word.
16. The system of claim 9 , wherein the instructions further cause the processor to train the neural network when the second processor indicates that the letter is correct or incorrect.
17. A method comprising:
executing, by a processor, an image recognition protocol purported to identify a letter within a word; in the event the image recognition protocol fails to identify the letter within the word, executing, by the processor, a neural network comprising a nodal data structure that identifies the letter within the word based upon a preceding letter within the word depicted within a check image, wherein the word is depicted within an image segment of the check image associated with a data field,
the nodal data structure comprising a set of nodes where each node represents a letter, each node within the set of nodes connected to at least one other node within the set of nodes, wherein connection of a first node to a second node corresponds to a probability that a letter corresponding to the second node is used subsequent to a preceding letter corresponding to the first node, wherein the processor queries an n-gram database to retrieve the probability of the letter being used after the preceding letter;
transmitting, by the processor, an identification of the letter to a second processor.
18. The method of claim 17 , wherein the word is depicted within predefined coordinates of the image.