Display screen quality detection method, apparatus, electronic device and storage medium

A display screen quality detection apparatus, wherein the display screen quality detection apparatus comprises:

• a processor, a memory, and a computer program stored on the memory and operable on the processor, and the processor executes the computer program to:
• receive a quality detection request sent by a console deployed on a display screen production line, wherein the quality detection request comprises a display screen image captured by an image capturing device on the display screen production line;
• perform image preprocessing on the display screen image, and input the preprocessed display screen image into a defect detection model to obtain a defect detection result, wherein the image preprocessing comprises one or more of following processes:
• trimming, cutting, rotating, reducing, and zooming, and wherein the defect detection model is obtained by training with a historical defect display screen image using a deep convolutional neural network structure and an instance segmentation algorithm;
• and
• determine, according to the defect detection result, quality of a display screen corresponding to the display screen image;
• wherein the defect detection model is a result obtained by performing combined training with candidate region loss, region category loss, region boundary loss, and pixel instance loss of the historical defect display screen image, to make a total loss value of the candidate region loss, the region category loss, the region boundary loss and the pixel instance loss meet a preset loss threshold;
• wherein the candidate region loss refers to a loss value between a selected defect region and an actual defect region in the historical defect display screen image, the region category loss refers to a loss value between a predicted defect category and an actual defect category in the selected defect region, the region boundary loss refers to a loss between a predicted defect boundary and an actual defect boundary in the selected defect region, and the pixel instance loss refers to a loss between a predicted pixel instance and an actual pixel instance in the historical defect display screen image.

A non-transitory storage medium, wherein the non-transitory storage medium stores instructions that, when run on a computer, cause the computer to perform following steps:

• receiving a quality detection request sent by a console deployed on a display screen production line, wherein the quality detection request comprises a display screen image captured by an image capturing device on the display screen production line;
• performing image preprocessing on the display screen image, wherein the image preprocessing comprises one or more of following processes:
• trimming, cutting, rotating, reducing, and zooming;
• inputting the preprocessed display screen image into a defect detection model to obtain a defect detection result, wherein the defect detection model is obtained by training with a historical defect display screen image using a deep convolutional neural network structure and an instance segmentation algorithm;
• and
• determining, according to the defect detection result, quality of a display screen corresponding to the display screen image;
• wherein the defect detection model is a result obtained by performing combined training with candidate region loss, region category loss, region boundary loss, and pixel instance loss of the historical defect display screen image to make a total loss value of the candidate region loss, the region category loss, the region boundary loss, and the pixel instance loss meet a preset loss threshold;
• wherein the candidate region loss refers to a loss value between a selected defect region and an actual defect region in the historical defect display screen image, the region category loss refers to a loss value between a predicted defect category and an actual defect category in the selected defect region, the region boundary loss refers to a loss between a predicted defect boundary and an actual defect boundary in the selected defect region, and the pixel instance loss refers to a loss between a predicted pixel instance and an actual pixel instance in the historical defect display screen image.