Apparatus for processing substrate and method of determining whether substrate treatment process is normal
The inventive concept provides a method to determine whether a substrate treatment process is normal using a deep learning model. The method comprising receiving input on a substrate treatment process video, preprocessing the inputted video, using the deep learning model to study a preprocessed video, and determining whether the substrate treatment process is normal by comparing the trained model and a real time substrate treatment process video.
1. A method for determining whether a substrate treatment process is normal using a deep learning model, the method comprising:
receiving an input image of a substrate processing;
preprocessing the input image;
choosing a recognized tip area of a nozzle as a region of interest (ROI);
learning the preprocessed input image by using a deep learning model; and
comparing a real time substrate treatment process image and a learned data by using the trained deep learning model to determine whether the substrate treatment process is normal,
wherein the preprocessing the input image comprises recognizing the tip area of a nozzle in the input image of the substrate processing, a chemical being sprayed from the tip area of the nozzle from which a chemical is to be sprayed,
wherein the learning the preprocessed input image by using the deep learning model comprises learning the ROI according to flow rates of the chemical.
2. The method of claim 1 , wherein the preprocessing the input image comprises correcting a perspective distortion of the input image.
3. The method of claim 2 , wherein the preprocessing the input image further comprises detecting a shape of a substrate from the perspective distortion corrected input image and detecting a central point of the shape of the substrate.
4. The method of claim 3 , wherein the preprocessing the input image further comprises converting a coordinate system to a polar coordinate system based on the detected central point.
5. The method of claim 4 further comprising choosing a constant edge range of the substrate in the polar coordinate system as a region of interest (ROI).
6. The method of claim 5 , wherein learning the preprocessed input image by using the deep learning model comprises learning the ROI.
7. A substrate treatment process apparatus for processing a substrate by spraying a chemical, the substrate treatment process apparatus comprising:
an image capturing unit for capturing an image of the substrate and a chemical being sprayed on the substrate; and
a determining unit for determining whether a substrate treatment process is normal by using a deep learning model, wherein the determining unit comprises:
a preprocessing unit for preprocessing data of the image captured by the image capturing unit;
a learning unit for learning the preprocessed data by using a deep learning model; and
a comparing unit that determines whether the substrate treatment process is normal by comparing data of a real time substrate treatment process image and the learned data in the learning unit,
wherein the determining unit determines whether the substrate treatment process is normal based on either an aerosol spray form inside a chamber in which the substrate processes is performed or a wetting of the substrate,
wherein the preprocessing unit chooses a part of the image of the substrate and a chemical sprayed on the substrate captured by the image capturing unit as a region of interest (ROI),
wherein the ROI is a tip area in an image of the substrate processing, the tip area being a chemical sprayed area by a nozzle, and
wherein the learning unit learns the ROI according to flow rates of the chemical.
8. The substrate treatment process apparatus of claim 7 , wherein the preprocessing unit is further configured to compensate a perspective distortion of the date of the mage.
9. The substrate treatment process apparatus of claim 8 , wherein the preprocessing unit is further configured to detect a shape of a substrate from the perspective distance distortion corrected data, and detect a central point from the shape of the substrate.
10. The substrate treatment process apparatus of claim 9 , wherein the preprocessing unit converts a coordinate system into a polar coordinate system based on the detected central point.
11. The substrate treatment process apparatus of claim 10 , wherein the preprocessing unit chooses a constant edge range of the substrate in the polar coordinate system as a ROI.
12. A method for determining whether a wetting state of a substrate by using a deep learning model, the method comprising:
receiving an input image of a substrate processing;
preprocessing the input image;
learning the preprocessed input image by using a deep learning model; and
comparing a real time substrate treatment process image and the learned data by the deep learning model and determining whether a substrate treatment process is normal,
wherein the preprocessing an input image comprises recognizing a tip area in the input image of the substrate processing, the tip area being a chemical sprayed area by a nozzle,
wherein the learning the preprocessed input image by using the deep learning model comprises learning the ROI according to flow rates of the chemical.