CN114580515A - Neural network training method for intelligent detection of semiconductor desoldering - Google Patents
Neural network training method for intelligent detection of semiconductor desoldering Download PDFInfo
- Publication number
- CN114580515A CN114580515A CN202210183221.1A CN202210183221A CN114580515A CN 114580515 A CN114580515 A CN 114580515A CN 202210183221 A CN202210183221 A CN 202210183221A CN 114580515 A CN114580515 A CN 114580515A
- Authority
- CN
- China
- Prior art keywords
- semiconductor
- neural network
- desoldering
- feature map
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30152—Solder
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The application relates to intelligent desoldering detection in the field of semiconductors, and particularly discloses a neural network training method for intelligent semiconductor desoldering detection. Specifically, a training image after semiconductor welding is classified in a high-dimensional space to obtain a classification loss function value, then an image qualified for desoldering detection is used as a reference image, and a feature difference of a region of interest of the reference image and the training image in the high-dimensional space is obtained to obtain a distance loss function value, and then a neural network is trained based on a weighted sum of the distance loss function value and the classification loss function value to enhance the training speed and precision.
Description
Technical Field
The present invention relates to intelligent desoldering detection in the semiconductor field, and more particularly, to a neural network training method for semiconductor desoldering intelligent detection, a semiconductor desoldering intelligent detection method based on a deep neural network, a neural network training system for semiconductor desoldering intelligent detection, a semiconductor desoldering intelligent detection system based on a deep neural network, and an electronic device.
Background
The copper wire bonding technology is a chip packaging technology which is relatively short in application time, not mature in technology but higher in market share in the current semiconductor manufacturing process. With the continuous improvement of various technical indexes of lead welding, the traditional gold wire and aluminum wire gradually tend to the index limit. The copper wire has good mechanical, electrical and thermal properties in chip lead welding, the copper wire replaces a gold wire and an aluminum wire to shorten the welding distance and improve the frequency and the reliability of a chip, but the weldability of the copper wire is poor due to the fact that the surface of the copper wire is easy to oxidize, a big problem is continuously exposed along with the continuous improvement of the lead welding speed, the problem is eutectic failure (desoldering defect) of a welding molten ball and a welding pad, when a eutectic surface is in failure critical, a device can pass an instant test, but the reliability of the device is poor. If the reliability of the device is poor, the phenomenon that a client generates clearance failure or fatigue failure at higher temperature in the using process can occur.
At present, a detection method for semiconductor desoldering comprises a visual detection method and a current detection method, however, the visual detection method requires an experienced worker to perform detection with the help of a high magnification magnifier, so that the labor cost is increased, meanwhile, misjudgment is easy to occur, and the current detection method is troublesome to operate and consumes time and labor.
Therefore, a technical solution capable of intelligently detecting the problem of the semiconductor debonding is desired.
At present, deep learning and neural networks have been widely applied in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks also exhibit a level close to or even exceeding that of humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
The deep learning and the development of the neural network provide a new solution for the intelligent detection of the semiconductor desoldering problem.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a neural network training method for semiconductor desoldering intelligent detection, a semiconductor desoldering intelligent detection method based on a deep neural network, a neural network training system for semiconductor desoldering intelligent detection, a semiconductor desoldering intelligent detection system based on a deep neural network and electronic equipment. Specifically, a training image after semiconductor welding is classified in a high-dimensional space to obtain a classification loss function value, then an image qualified for desoldering detection is used as a reference image, and a feature difference of a region of interest of the reference image and the training image in the high-dimensional space is obtained to obtain a distance loss function value, and then a neural network is trained based on a weighted sum of the distance loss function value and the classification loss function value to enhance the training speed and precision.
According to one aspect of the application, a neural network training method for semiconductor desoldering intelligent detection is provided, which comprises the following steps:
acquiring a high-precision training image of a welded semiconductor;
passing the training image through a first convolutional neural network to extract a training feature map corresponding to the training image;
enabling the reference image to pass through a second convolutional neural network so as to extract a reference characteristic map corresponding to the reference image, wherein the reference image is an image qualified in semiconductor welding;
amplifying the training feature map and the reference feature map by n times to obtain a first amplified feature map corresponding to the training feature map and a second amplified feature map corresponding to the reference feature map;
acquiring a plurality of first regions of interest in the first magnified feature map and a plurality of second regions of interest in the second magnified feature map;
calculating distance loss functions between corresponding regions of interest to obtain a plurality of distance loss function values;
passing the training feature map through a classifier to obtain a classification loss function value; and
updating parameters of the first convolutional neural network and the classifier based on a weighted sum of the distance loss function values and the classification loss function values.
In the above neural network training method for semiconductor desoldering intelligent detection, the first convolutional neural network and the second convolutional neural network have the same network structure, and the first convolutional neural network and the second convolutional neural network are deep residual error networks.
In the above neural network training method for semiconductor desoldering intelligent detection, n is a positive integer greater than or equal to 5.
In the above neural network training method for semiconductor desoldering intelligent detection, acquiring a plurality of first regions of interest in a first enlarged feature map and a plurality of second regions of interest in a second enlarged feature map includes: determining the positions of welding pins in the first enlarged characteristic diagram and the second enlarged characteristic diagram; and enclosing the welding pins by using the rectangular candidate frame and amplifying by taking the positions of the welding pins as centers to obtain an interested area containing the surrounding areas of the pins.
In the above neural network training method for semiconductor desoldering intelligent detection, calculating a distance loss function between corresponding regions of interest to obtain a plurality of distance loss function values includes: and calculating Euclidean distances between corresponding interested areas to obtain a plurality of distance loss function values.
In the above neural network training method for semiconductor desoldering intelligent detection, calculating a distance loss function between corresponding regions of interest to obtain a plurality of distance loss function values, includes: cosine distances between the corresponding regions of interest are calculated to obtain a plurality of distance loss function values.
According to another aspect of the present application, there is provided a semiconductor desoldering intelligent detection method based on a deep neural network, which includes:
acquiring a high-precision image of a welded semiconductor to be detected;
inputting the image into a first convolution neural network and a classifier which are trained according to the neural network training method for the intelligent detection of the semiconductor desoldering, wherein the output of the classifier represents a first probability corresponding to the semiconductor desoldering detection being qualified and a second probability corresponding to the semiconductor desoldering detection being unqualified; and
and determining whether the semiconductor desoldering detection is qualified or not based on the first probability and the second probability.
According to yet another aspect of the present application, there is provided a neural network training system for semiconductor desoldering intelligent detection, comprising:
a training image acquisition unit for acquiring a high-precision training image of the semiconductor after welding;
the training feature map generating unit is used for enabling the training image to pass through a first convolutional neural network so as to extract a training feature map corresponding to the training image;
the reference characteristic map generating unit is used for enabling a reference image to pass through a second convolutional neural network so as to extract a reference characteristic map corresponding to the reference image, wherein the reference image is an image qualified for semiconductor welding;
an enlarged feature map generating unit configured to enlarge the training feature map obtained by the training feature map generating unit and the reference feature map obtained by the reference feature map generating unit by n times to obtain a first enlarged feature map corresponding to the training feature map and a second enlarged feature map corresponding to the reference feature map;
a region-of-interest generating unit configured to acquire a plurality of first regions of interest in the first enlarged feature map and a plurality of second regions of interest in the second enlarged feature map acquired by the enlarged feature map generating unit;
a distance loss function value generating unit configured to calculate a distance loss function between corresponding regions of interest among the plurality of regions of interest obtained by the region of interest generating unit to obtain a plurality of distance loss function values;
the classification loss function value generating unit is used for enabling the training feature map obtained by the training feature map generating unit to pass through a classifier so as to obtain a classification loss function value;
a parameter updating unit configured to update parameters of the first convolutional neural network and the classifier based on a weighted sum of the distance loss function value obtained by the distance loss function value generating unit and the classification loss function value obtained by the classification loss function value generating unit.
In the above neural network training system for semiconductor desoldering intelligent detection, the first convolutional neural network and the second convolutional neural network have the same network structure, and the first convolutional neural network and the second convolutional neural network are deep residual error networks.
In the above neural network training system for semiconductor desoldering intelligent detection, n is a positive integer greater than or equal to 5.
In the above neural network training system for semiconductor desoldering intelligent detection, the region of interest generating unit includes: the position determining subunit is used for determining the positions of the welding pins in the first amplified characteristic diagram and the second amplified characteristic diagram; and the amplifying subunit is used for determining the positions of the welding pins obtained by the subunit by surrounding the positions by the rectangular candidate frame and amplifying by taking the positions of the welding pins as the center so as to obtain an interested area containing the surrounding areas of the pins.
In the above neural network training system for semiconductor desoldering intelligent detection, the distance loss function value generating unit is further configured to: and calculating Euclidean distances between corresponding interested areas to obtain a plurality of distance loss function values.
In the above neural network training system for semiconductor desoldering intelligent detection, the distance loss function value generating unit is further configured to: cosine distances between corresponding regions of interest are calculated to obtain a plurality of distance loss function values.
According to another aspect of the present application, there is provided a semiconductor desoldering intelligent detection system based on a deep neural network, including:
the device comprises an image acquisition unit to be detected, a data acquisition unit and a data processing unit, wherein the image acquisition unit to be detected is used for acquiring a high-precision image of a welded semiconductor to be detected;
the classification unit is used for inputting the image to be detected obtained by the image to be detected acquisition unit into a first convolution neural network and a classifier which are trained according to the neural network training method for the intelligent semiconductor desoldering detection, and the output of the classifier represents a first probability corresponding to the semiconductor desoldering detection being qualified and a second probability corresponding to the semiconductor desoldering detection being unqualified; and
and the detection result generation unit is used for generating a detection result of whether the semiconductor desoldering detection is qualified or not based on the first probability and the second probability obtained by the classification unit.
According to still another aspect of the present application, there is provided an electronic apparatus including: a processor; and a memory having stored therein computer program instructions which, when executed by the processor, cause the processor to perform a neural network training method for semiconductor desoldering intelligence testing, or a semiconductor desoldering intelligence testing method based on a deep neural network, as described above.
According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to execute a neural network training method for semiconductor desoldering intelligent detection or a semiconductor desoldering intelligent detection method based on a deep neural network as described above.
Compared with the prior art, the neural network training method for semiconductor desoldering intelligent detection, the semiconductor desoldering intelligent detection method based on the deep neural network, the neural network training system for semiconductor desoldering intelligent detection, the semiconductor desoldering intelligent detection system based on the deep neural network and the electronic equipment provided by the application have the advantages that the obtained image after semiconductor desoldering is identified based on the deep learning computer vision method, and the detection result of whether the semiconductor desoldering detection is qualified or not is obtained based on the classification mode of the identified high-dimensional image features. Specifically, a training image after semiconductor welding is classified in a high-dimensional space to obtain a classification loss function value, then an image qualified for desoldering detection is used as a reference image, and a feature difference of a region of interest of the reference image and the training image in the high-dimensional space is obtained to obtain a distance loss function value, and then a neural network is trained based on a weighted sum of the distance loss function value and the classification loss function value to enhance the training speed and precision.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
FIG. 1 illustrates a flow diagram of a neural network training method for intelligent detection of semiconductor desoldering according to an embodiment of the present application;
FIG. 2 illustrates a system architecture diagram of a neural network training method for semiconductor desoldering intelligence detection in accordance with an embodiment of the present application;
FIG. 3 illustrates a flow chart for obtaining a plurality of first regions of interest in a first enlarged feature map and a plurality of second regions of interest in a second enlarged feature map in a neural network training method for intelligent detection of semiconductor desoldering according to an embodiment of the present application;
FIG. 4 illustrates a flow diagram of a deep neural network based semiconductor desoldering intelligent detection method according to an embodiment of the present application;
FIG. 5 illustrates a block diagram of a neural network training system for semiconductor desoldering intelligence detection in accordance with an embodiment of the present application;
FIG. 6 illustrates a block diagram of a region of interest generating unit in a neural network training system for intelligent detection of semiconductor desoldering according to an embodiment of the present application;
FIG. 7 illustrates a block diagram of a deep neural network based semiconductor desoldering intelligent detection system in accordance with an embodiment of the present application;
FIG. 8 illustrates a block diagram of an electronic device in accordance with an embodiment of the application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of a scene
As mentioned above, copper has poor solderability due to surface oxidation, and as the wire bonding speed is increased, a big problem, namely eutectic failure (debonding defect) between the bonding ball and the bonding pad, is exposed continuously, and when the eutectic surface is at the failure critical point, the device can pass the transient test, but the reliability of the device is poor. If the reliability of the device is poor, the phenomenon that a client generates clearance failure or fatigue failure at higher temperature in the using process can occur.
At present, a detection method for semiconductor desoldering comprises a visual detection method and a current detection method, however, the visual detection method requires an experienced worker to perform detection with the help of a high magnification magnifier, so that the labor cost is increased, meanwhile, misjudgment is easy to occur, and the current detection method is troublesome to operate and consumes time and labor. Therefore, the inventors of the present application expect to be able to propose a method of intelligently detecting a problem of semiconductor desoldering.
In practice, the inventor of the present application finds that the soldered pins are generally very smooth. When the edge is affected, the edge becomes rough and dull due to continuous extrusion and stretching, a dark circle appears around a welding point, a crack-shaped fine crack group can be seen by a high magnification magnifier, and an annular crack, namely, desoldering, is formed in serious cases. Therefore, the area with the ring-shaped black circle is a hidden danger in the future even if no desoldering exists.
Based on this, the inventors of the present application consider that the image after the semiconductor bonding is subjected to feature extraction and recognition by a computer vision method based on deep learning, thereby converting the detection of the semiconductor debonding into a classification problem based on the image.
Specifically, in the scheme of the application, a high-precision image after the semiconductor welding is completed is firstly obtained as a training image, and then the training image passes through a first convolutional neural network to extract high-dimensional features in the training image through the convolutional neural network so as to obtain a training feature map. And then, taking the image qualified by the desoldering detection as a reference image, and mapping the reference image into the image high-dimensional space through a second convolution neural network with the same structure to obtain a reference feature map. Since the region concerned by the problem of the desoldering is the region where the semiconductor bonding pin is located, it is necessary to obtain a region of interest in the training image, that is, a rectangular region centered on the bonding pin is selected as the region of interest through the rectangular candidate box. And then calculating the difference of the interested areas of the training image and the reference image in a high-dimensional space to obtain a distance loss function value. Meanwhile, the training characteristic diagram is classified through a classifier, cross entropy loss function values between the classification results and real values are calculated to obtain classification loss function values, then the weighted sum of the distance loss function values and the classification loss function values is calculated, the weighted sum is reduced by a preset step length to train the first convolution neural network and the classifier, so that the first convolution neural network learns the characteristics of whether the semiconductor is subjected to off-welding classification or not, and the training speed and the training precision of the model are improved.
Based on this, the present application proposes a neural network training method for semiconductor desoldering intelligent detection, which includes: acquiring a high-precision training image of a welded semiconductor; passing the training image through a first convolutional neural network to extract a training feature map corresponding to the training image; enabling the reference image to pass through a second convolutional neural network so as to extract a reference characteristic map corresponding to the reference image, wherein the reference image is an image qualified in semiconductor welding; amplifying the training feature map and the reference feature map by n times to obtain a first amplified feature map corresponding to the training feature map and a second amplified feature map corresponding to the reference feature map; acquiring a plurality of first interested areas in the first enlarged feature map and a plurality of second interested areas in the second enlarged feature map; calculating distance loss functions between corresponding regions of interest to obtain a plurality of distance loss function values; passing the training feature map through a classifier to obtain a classification loss function value; and updating parameters of the first convolutional neural network and the classifier based on a weighted sum of the distance loss function values and the classification loss function values.
Based on the above, the present application further provides a semiconductor desoldering intelligent detection method based on a deep neural network, which includes: acquiring a high-precision image of a welded semiconductor to be detected; inputting the image into a first convolution neural network and a classifier which are trained according to the neural network training method for the intelligent detection of the semiconductor desoldering, wherein the output of the classifier represents a first probability corresponding to the semiconductor desoldering detection being qualified and a second probability corresponding to the semiconductor desoldering detection being unqualified; and determining whether the semiconductor desoldering detection is qualified or not based on the first probability and the second probability.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary method
FIG. 1 illustrates a flow diagram of a neural network training method for semiconductor desoldering intelligence detection. As shown in fig. 1, a neural network training method for semiconductor desoldering intelligent detection according to an embodiment of the present application includes: s110, acquiring a high-precision training image of the welded semiconductor; s120, enabling the training image to pass through a first convolutional neural network to extract a training feature map corresponding to the training image; s130, enabling the reference image to pass through a second convolutional neural network so as to extract a reference characteristic diagram corresponding to the reference image, wherein the reference image is an image qualified for semiconductor welding; s140, amplifying the training feature map and the reference feature map by n times to obtain a first amplified feature map corresponding to the training feature map and a second amplified feature map corresponding to the reference feature map; s150, acquiring a plurality of first interested areas in the first amplified characteristic diagram and a plurality of second interested areas in the second amplified characteristic diagram; s160, calculating distance loss functions between corresponding interested areas to obtain a plurality of distance loss function values; s170, passing the training feature map through a classifier to obtain a classification loss function value; and S180, updating parameters of the first convolutional neural network and the classifier based on the weighted sum of the distance loss function values and the classification loss function values.
Fig. 2 illustrates an architectural schematic diagram of a neural network training method for semiconductor desoldering intelligence detection according to an embodiment of the present application. As shown IN fig. 2, IN the network architecture of the neural network training method for semiconductor desoldering intelligent detection, first, an acquired training image (e.g., IN1 as illustrated IN fig. 2) is passed through a first convolutional neural network (e.g., CNN1 as illustrated IN fig. 2) to obtain a training feature map (e.g., F1 as illustrated IN fig. 2); next, passing a reference image (e.g., IN2 as illustrated IN fig. 2) through a second convolutional neural network (e.g., CNN2 as illustrated IN fig. 2) to extract a reference feature map (e.g., F2 as illustrated IN fig. 2) corresponding to the reference image; then, the training and reference signatures are enlarged by a factor of n to obtain a first enlarged signature (e.g., Fa1 as illustrated in fig. 2) corresponding to the training signature and a second enlarged signature (e.g., Fa2 as illustrated in fig. 2) corresponding to the reference signature; next, acquiring a plurality of first regions of interest in the first magnified rendering (e.g., D1 as illustrated in fig. 2) and a plurality of second regions of interest in the second magnified rendering (e.g., D2 as illustrated in fig. 2); then, calculating distance loss functions between corresponding interested areas to obtain a plurality of distance loss function values; then, passing the training feature map through a classifier (e.g., a classifier as illustrated in fig. 2) to obtain a classification loss function value; then, parameters of the first convolutional neural network and the classifier are updated based on a weighted sum of the distance loss function values and the classification loss function values.
In step S110, a high-precision training image of the soldered semiconductor is acquired. As described above, the present invention considers that the obtained semiconductor image after soldering is recognized, and the detection result of whether the semiconductor desoldering detection is acceptable or not is obtained by classifying the semiconductor image based on the recognized high-dimensional image features. Specifically, in the embodiment of the present application, a high-precision image of a semiconductor subjected to soldering is first acquired by a camera as a training image.
In step S120, the training image is passed through a first convolutional neural network to extract a training feature map corresponding to the training image. That is, the high-dimensional features in the training image are extracted with a first convolutional neural network.
In particular, the first convolutional neural network may employ a deep residual neural network, e.g., ResNet 50. It should be known to those skilled in the art that, compared to the conventional convolutional neural network, the deep residual network is an optimized network structure proposed on the basis of the conventional convolutional neural network, which mainly solves the problem of gradient disappearance during the training process. The depth residual error network introduces a residual error network structure, the network layer can be made deeper through the residual error network structure, and the problem of gradient disappearance can not occur. The residual error network uses the cross-layer link thought of a high-speed network for reference, breaks through the convention that the traditional neural network only can provide N layers as input from the input layer of the N-1 layer, enables the output of a certain layer to directly cross several layers as the input of the later layer, and has the significance of providing a new direction for the difficult problem that the error rate of the whole learning model is not reduced and inversely increased by superposing multiple layers of networks.
In step S130, the reference image is passed through a second convolutional neural network to extract a reference feature map corresponding to the reference image, wherein the reference image is an image qualified for semiconductor welding. That is, each high-dimensional feature in the reference image is extracted with a second convolutional neural network.
Specifically, in the embodiment of the present application, the first convolutional neural network and the second convolutional neural network have the same network structure, that is, the weights of the first convolutional neural network and the second convolutional neural network are partially shared, so as to reduce the amount of computation in the training process, which is beneficial to avoiding the disappearance of the gradient. Meanwhile, the training characteristic diagrams output by the first convolutional neural network and the second convolutional neural network with the same network structure have the same scale as the reference characteristic diagram, so that subsequent calculation is facilitated. In particular, the second convolutional neural network is a deep residual network, e.g., ResNet 50.
In step S140, the training feature map and the reference feature map are enlarged by n times to obtain a first enlarged feature map corresponding to the training feature map and a second enlarged feature map corresponding to the reference feature map.
Specifically, in the embodiment of the present application, n is a positive integer greater than or equal to 5. It should be understood that, because the soldering area of the pins is small in the training image of the soldering semiconductor, an enlarged characteristic diagram convenient for comparison and calculation is obtained by acquiring a high-precision image of the soldering semiconductor and then enlarging the image.
In step S150, a plurality of first regions of interest in the first magnified feature map and a plurality of second regions of interest in the second magnified feature map are acquired. It should be understood that in the semiconductor desoldering inspection, the positions of the regions of interest, i.e., the regions of interest, in the first enlarged feature map and the second enlarged feature map need to be acquired.
Specifically, in the embodiment of the present application, the process of acquiring a plurality of first regions of interest in a first enlarged feature map and a plurality of second regions of interest in a second enlarged feature map includes: first, the positions of the bonding pins in the first and second enlarged characteristic maps are determined, and it is understood that in the semiconductor desoldering inspection, the criterion related to the bonding quality is the quality of the bonding pins, and thus a high-dimensional feature of an image of the position region of the bonding pins needs to be focused. Then, the soldering pin is enclosed by a rectangular candidate frame and enlarged with the position of the soldering pin as the center to obtain an area of interest including the area around the pin, that is, a rectangular area with the soldering pin as the center is selected by the rectangular candidate frame to obtain a plurality of areas of interest. One of ordinary skill in the art will appreciate that in the field of image processing, a region of interest (ROI) is an image region selected from an image, which is the focus of your image analysis. The area is delineated for further processing. The ROI is used for delineating the target which the user wants to read, so that the processing time can be reduced, and the precision can be increased.
Fig. 3 illustrates a flowchart for acquiring a plurality of first regions of interest in a first enlarged feature map and a plurality of second regions of interest in a second enlarged feature map in a neural network training method for semiconductor desoldering intelligence detection according to an embodiment of the present application. As shown in fig. 4, acquiring a plurality of first regions of interest in the first magnified rendering and a plurality of second regions of interest in the second magnified rendering includes: s210, determining the positions of the welding pins in the first enlarged characteristic diagram and the second enlarged characteristic diagram; and S220, enclosing the welding pin by a rectangular candidate frame and amplifying by taking the position of the welding pin as a center to obtain an interested area containing the surrounding area of the pin.
In step S160, distance loss functions between corresponding regions of interest are calculated to obtain a plurality of distance loss function values. That is, a distance loss function between a plurality of first regions of interest in the first enlarged feature map and a plurality of second regions of interest in the second enlarged feature map is calculated to obtain a plurality of distance loss function values.
Specifically, in the embodiment of the present application, a process of calculating a distance loss function between corresponding regions of interest to obtain a plurality of distance loss function values includes: and calculating Euclidean distances between corresponding interested areas to obtain a plurality of distance loss function values. As will be appreciated by those skilled in the art, the euclidean distance is the sum of the squares of the differences between the target values and the estimated values, and by calculating the euclidean distance between corresponding regions of interest in the first and second magnified feature maps, the feature differences between each position of each two corresponding regions of interest in the plurality of regions of interest can be reflected in the spatial distance dimension.
In particular, in the embodiment of the present application, the distance loss function between corresponding regions of interest may also be calculated by other methods to obtain a plurality of distance loss function values. For example, in other examples of the present application, a process of calculating a distance loss function between corresponding regions of interest to obtain a plurality of distance loss function values includes: cosine distances between the corresponding regions of interest are calculated to obtain a plurality of distance loss function values. It should be understood that the similarity between the corresponding regions of interest can be analyzed by calculating the cosine distance between the corresponding regions of interest, so that the first convolutional neural network and the classifier are trained by minimizing the distance loss function, so that the first convolutional neural network can extract the image features close to the reference feature map in the training feature map.
In step S170, the training feature map is passed through a classifier to obtain a classification loss function value. That is, in a decoupled manner, the classifier contains an encoder that may be composed of convolutional layers, pooled layers, or fully-connected layers.
Specifically, in the embodiment of the present application, the process of passing the training feature map through a classifier to obtain a classification loss function value includes: first, the training feature map is passed through one or more fully-connected layers to obtain a classification feature vector, that is, the training feature map is encoded by using the one or more fully-connected layers as an encoder to generate a classification feature vector. Then, the classification feature vector is input into a Softmax classification function to obtain a classification result. Then, the classification result and the real value are input into a loss function to obtain the classification loss function value.
In step S180, parameters of the first convolutional neural network and the classifier are updated based on a weighted sum of the distance loss function values and the classification loss function values. It should be appreciated that training the first convolutional neural network based on the weighted sum of the classification loss function values and the distance loss function values, and updating the parameters of the first convolutional neural network accordingly, can enable the first convolutional neural network to learn how to perform high-dimensional feature extraction on images with respect to high-dimensional features of the images, and can enhance the training speed and precision of the model.
According to another aspect of the application, a semiconductor desoldering intelligent detection method based on a deep neural network is further provided.
Fig. 4 illustrates a flowchart of a deep neural network-based semiconductor desoldering intelligent detection method according to an embodiment of the present application. As shown in fig. 4, the intelligent semiconductor desoldering detection method based on the deep neural network according to the embodiment of the present application includes: s310, acquiring a high-precision image of the welded semiconductor to be detected; s320, inputting the image into a first convolution neural network and a classifier trained according to the neural network training method for the intelligent semiconductor desoldering detection, wherein the output of the classifier represents a first probability corresponding to the semiconductor desoldering detection being qualified and a second probability corresponding to the semiconductor desoldering detection being unqualified; and S330, determining whether the semiconductor desoldering detection is qualified or not based on the first probability and the second probability.
In summary, the neural network training method for semiconductor desoldering intelligent detection and the semiconductor desoldering intelligent detection method based on the deep neural network are clarified, and the method based on the deep learning computer vision is used for identifying the obtained semiconductor welded images and classifying the images based on the identified high-dimensional image characteristics to obtain the detection result whether the semiconductor desoldering detection is qualified or not. Specifically, a training image after semiconductor welding is classified in a high-dimensional space to obtain a classification loss function value, then an image qualified for desoldering detection is used as a reference image, and a feature difference of a region of interest of the reference image and the training image in the high-dimensional space is obtained to obtain a distance loss function value, and then a neural network is trained based on a weighted sum of the distance loss function value and the classification loss function value to enhance the training speed and precision.
Exemplary System
FIG. 5 illustrates a block diagram of a neural network training system for semiconductor desoldering intelligence detection in accordance with an embodiment of the present application.
As shown in fig. 5, a neural network training system 500 for semiconductor desoldering intelligent detection according to an embodiment of the present application includes: a training image acquisition unit 510 for acquiring a high-precision training image of the semiconductor after the soldering is completed; a training feature map generating unit 520, configured to pass the training image obtained by the training image obtaining unit 510 through a first convolutional neural network to extract a training feature map corresponding to the training image; a reference feature map generating unit 530, configured to pass a reference image through a second convolutional neural network to extract a reference feature map corresponding to the reference image, where the reference image is an image that is qualified for semiconductor welding; an enlarged feature map generating unit 540, configured to enlarge the training feature map obtained by the training feature map generating unit 520 and the reference feature map obtained by the reference feature map generating unit 530 by n times, so as to obtain a first enlarged feature map corresponding to the training feature map and a second enlarged feature map corresponding to the reference feature map; a region-of-interest generating unit 550, configured to acquire a plurality of first regions of interest in the first enlarged feature map and a plurality of second regions of interest in the second enlarged feature map, where the plurality of first regions of interest are acquired by the enlarged feature map generating unit 540; a distance loss function value generating unit 560 for calculating a distance loss function between corresponding regions of interest among the plurality of regions of interest obtained by the region of interest generating unit 550 to obtain a plurality of distance loss function values; a classification loss function value generating unit 570, configured to pass the training feature map obtained by the training feature map generating unit 520 through a classifier to obtain a classification loss function value; and a parameter updating unit 580 for updating parameters of the first convolutional neural network and the classifier based on a weighted sum of the distance loss function value obtained by the distance loss function value generating unit 560 and the classification loss function value obtained by the classification loss function value generating unit 570.
In one example, in the above neural network training system 500 for semiconductor desoldering intelligent detection, the first convolutional neural network and the second convolutional neural network have the same network structure, and the first convolutional neural network and the second convolutional neural network are depth residual error networks.
In one example, in the above neural network training system 500 for semiconductor desoldering intelligent detection, n is a positive integer greater than or equal to 5.
In an example, in the above neural network training system 500 for intelligent detection of semiconductor desoldering, as shown in fig. 6, the region of interest generating unit 550 includes: a position determining subunit 551, configured to determine positions of the bonding pins in the first enlarged characteristic diagram and the second enlarged characteristic diagram; and an enlarging subunit 552, configured to determine the positions of the bonding pins obtained by the subunit 551 by surrounding the positions with rectangular candidate frames and enlarge with the bonding pin positions as the center to obtain a region of interest including the area around the pins.
In one example, in the above neural network training system 500 for semiconductor desoldering intelligent detection, the distance loss function value generating unit 560 is further configured to: and calculating Euclidean distances between corresponding interested areas to obtain a plurality of distance loss function values.
In one example, in the above neural network training system 500 for semiconductor desoldering intelligent detection, the distance loss function value generating unit 560 is further configured to: cosine distances between corresponding regions of interest are calculated to obtain a plurality of distance loss function values.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the training system 500 described above have been described in detail in the description of the neural network training method for semiconductor desoldering intelligence detection with reference to fig. 1 to 3, and thus, a repetitive description thereof will be omitted.
As described above, the training system 500 according to the embodiment of the present application may be implemented in various terminal devices, such as a server for semiconductor desoldering detection, and the like. In one example, the training system 500 according to embodiments of the present application may be integrated into the terminal device as a software module and/or a hardware module. For example, the training system 500 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the training system 500 may also be one of many hardware modules of the terminal device.
Alternatively, in another example, the training system 500 and the terminal device may be separate devices, and the training system 500 may be connected to the terminal device through a wired and/or wireless network and transmit the interactive information according to an agreed data format.
According to another aspect of the application, a semiconductor desoldering intelligent detection system based on a deep neural network is further provided.
Fig. 7 illustrates a block diagram of a deep neural network based semiconductor desoldering intelligent detection system according to an embodiment of the present application. As shown in fig. 7, the semiconductor desoldering intelligent detection system 700 based on the deep neural network according to the embodiment of the present application includes: an image-to-be-detected acquiring unit 710 for acquiring a high-precision image of a welded semiconductor to be detected; a classification unit 720, configured to input the image to be detected obtained by the image to be detected obtaining unit 710 into a first convolution neural network and a classifier trained according to the neural network training method for semiconductor desoldering intelligent detection as described above, where an output of the classifier represents a first probability corresponding to semiconductor desoldering detection qualification and a second probability corresponding to semiconductor desoldering detection disqualification; and a detection result generating unit 730 for generating a detection result of whether the semiconductor desoldering test is qualified or not based on the first probability and the second probability obtained by the classifying unit 720.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described smart detection system 700 have been described in detail in the above description of the deep neural network-based semiconductor desoldering smart detection method with reference to fig. 4, and thus, a repetitive description thereof will be omitted.
As described above, the smart detection system 700 according to the embodiment of the present application may be implemented in various terminal devices, such as a server for semiconductor desoldering detection, and the like. In one example, the smart detection system 700 according to embodiments of the present application may be integrated into a terminal device as a software module and/or a hardware module. For example, the smart detection system 700 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the smart detection system 700 could also be one of many hardware modules of the terminal device.
Alternatively, in another example, the smart detection system 700 and the terminal device may be separate devices, and the smart detection system 700 may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information according to an agreed data format.
Exemplary electronic device
Next, an electronic apparatus according to an embodiment of the present application is described with reference to fig. 8.
FIG. 8 illustrates a block diagram of an electronic device in accordance with an embodiment of the present application.
As shown in fig. 8, the electronic device comprises 10 at least one processor 11 and at least one memory 12.
The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
In one example, the electronic device 10 may further include: an input system 13 and an output system 14, which are interconnected by a bus system and/or other form of connection mechanism (not shown).
The input system 13 may comprise, for example, a keyboard, a mouse, etc.
The output system 14 can output various information including the detection result to the outside. The output system 14 may include, for example, a display, speakers, printer, and a communication network and its connected remote output devices, among others.
Of course, for simplicity, only some of the components of the electronic device 10 relevant to the present application are shown in fig. 8, and components such as buses, input/output interfaces, and the like are omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.
Exemplary computer program product and computer-readable storage Medium
In addition to the above-described methods and apparatus, embodiments of the present application may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the neural network training method for semiconductor desoldering intelligence testing, or the functions in the semiconductor desoldering intelligence testing method based on a deep neural network, according to various embodiments of the present application described in the "exemplary methods" section above in this specification.
The computer program product may be written with program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present application may also be a computer readable storage medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the neural network training method for semiconductor desoldering intelligence testing described in the "exemplary methods" section of this specification above, or the steps in the semiconductor desoldering intelligence testing method based on a deep neural network.
The computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the foregoing disclosure is not intended to be exhaustive or to limit the disclosure to the precise details disclosed.
The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.
Claims (10)
1. A neural network training method for intelligent detection of semiconductor desoldering is characterized by comprising the following steps:
acquiring a high-precision training image of a welded semiconductor;
passing the training image through a first convolutional neural network to extract a training feature map corresponding to the training image;
enabling the reference image to pass through a second convolutional neural network so as to extract a reference characteristic map corresponding to the reference image, wherein the reference image is an image qualified in semiconductor welding;
amplifying the training feature map and the reference feature map by n times to obtain a first amplified feature map corresponding to the training feature map and a second amplified feature map corresponding to the reference feature map;
acquiring a plurality of first interested areas in the first enlarged feature map and a plurality of second interested areas in the second enlarged feature map;
calculating distance loss functions between corresponding regions of interest to obtain a plurality of distance loss function values;
passing the training feature map through a classifier to obtain a classification loss function value; and
updating parameters of the first convolutional neural network and the classifier based on a weighted sum of the distance loss function values and the classification loss function values.
2. The neural network training method for semiconductor desoldering intelligent detection as recited in claim 1, wherein the first convolutional neural network and the second convolutional neural network have the same network structure, and the first convolutional neural network and the second convolutional neural network are deep residual error networks.
3. The neural network training method for semiconductor desoldering intelligent detection as recited in claim 1, wherein n is a positive integer greater than or equal to 5.
4. The neural network training method for semiconductor desoldering intelligence inspection as recited in claim 1, wherein obtaining a plurality of first regions of interest in a first magnified signature and a plurality of second regions of interest in a second magnified signature comprises:
determining the positions of welding pins in the first enlarged characteristic diagram and the second enlarged characteristic diagram; and
and enclosing the welding pins by using a rectangular candidate frame and amplifying by taking the positions of the welding pins as centers to obtain an interested area containing the surrounding areas of the pins.
5. The neural network training method for semiconductor desoldering intelligent detection as recited in claim 1, wherein computing distance loss functions between corresponding regions of interest to obtain a plurality of distance loss function values comprises:
and calculating Euclidean distances between corresponding interested areas to obtain a plurality of distance loss function values.
6. The neural network training method for intelligent detection of semiconductor desoldering according to claim 1, wherein calculating a distance loss function between corresponding regions of interest to obtain a plurality of distance loss function values comprises:
cosine distances between the corresponding regions of interest are calculated to obtain a plurality of distance loss function values.
7. A semiconductor desoldering intelligent detection method based on a deep neural network is characterized by comprising the following steps:
acquiring a high-precision image of a welded semiconductor to be detected;
inputting the image into a first convolutional neural network trained according to the neural network training method for the intelligent detection of the semiconductor desoldering as claimed in any one of claims 1 to 6 and a classifier, wherein the output of the classifier represents a first probability corresponding to the semiconductor desoldering detection being qualified and a second probability corresponding to the semiconductor desoldering detection being unqualified; and
and determining whether the semiconductor desoldering detection is qualified or not based on the first probability and the second probability.
8. A neural network training system for semiconductor desoldering intelligent detection, comprising:
a training image acquisition unit for acquiring a high-precision training image of the semiconductor after welding;
the training feature map generating unit is used for enabling the training image to pass through a first convolutional neural network so as to extract a training feature map corresponding to the training image;
the reference characteristic map generating unit is used for enabling a reference image to pass through a second convolutional neural network so as to extract a reference characteristic map corresponding to the reference image, wherein the reference image is an image qualified for semiconductor welding;
an enlarged feature map generating unit configured to enlarge the training feature map obtained by the training feature map generating unit and the reference feature map obtained by the reference feature map generating unit by n times to obtain a first enlarged feature map corresponding to the training feature map and a second enlarged feature map corresponding to the reference feature map;
a region-of-interest generating unit configured to acquire a plurality of first regions of interest in the first enlarged feature map and a plurality of second regions of interest in the second enlarged feature map acquired by the enlarged feature map generating unit;
a distance loss function value generating unit configured to calculate a distance loss function between corresponding regions of interest among the plurality of regions of interest obtained by the region of interest generating unit to obtain a plurality of distance loss function values;
the classification loss function value generating unit is used for enabling the training feature map obtained by the training feature map generating unit to pass through a classifier so as to obtain a classification loss function value;
a parameter updating unit configured to update parameters of the first convolutional neural network and the classifier based on a weighted sum of the distance loss function value obtained by the distance loss function value generating unit and the classification loss function value obtained by the classification loss function value generating unit.
9. A semiconductor desoldering intelligent detection system based on a deep neural network is characterized by comprising:
the device comprises an image acquisition unit to be detected, a data acquisition unit and a data processing unit, wherein the image acquisition unit to be detected is used for acquiring a high-precision image of a welded semiconductor to be detected;
the classification unit is used for inputting the image to be detected obtained by the image to be detected acquisition unit into a first convolution neural network and a classifier trained according to the neural network training method for the intelligent semiconductor desoldering detection as claimed in any one of claims 1 to 6, and the output of the classifier represents a first probability corresponding to the semiconductor desoldering detection being qualified and a second probability corresponding to the semiconductor desoldering detection being unqualified; and
and the detection result generating unit is used for generating a detection result of whether the semiconductor desoldering detection is qualified or not based on the first probability and the second probability obtained by the classifying unit.
10. An electronic device, comprising:
a processor; and
a memory having stored therein computer program instructions which, when executed by the processor, cause the processor to perform the neural network training method for semiconductor desoldering intelligence test of any of claims 1-6 or the deep neural network based semiconductor desoldering intelligence test method of claim 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210183221.1A CN114580515A (en) | 2022-02-28 | 2022-02-28 | Neural network training method for intelligent detection of semiconductor desoldering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210183221.1A CN114580515A (en) | 2022-02-28 | 2022-02-28 | Neural network training method for intelligent detection of semiconductor desoldering |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114580515A true CN114580515A (en) | 2022-06-03 |
Family
ID=81777128
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210183221.1A Withdrawn CN114580515A (en) | 2022-02-28 | 2022-02-28 | Neural network training method for intelligent detection of semiconductor desoldering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114580515A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114972327A (en) * | 2022-07-12 | 2022-08-30 | 爱尔达电气有限公司 | Semiconductor package test system and test method thereof |
| CN115375691A (en) * | 2022-10-26 | 2022-11-22 | 济宁九德半导体科技有限公司 | Image-based semiconductor diffusion paper source defect detection system and method thereof |
| CN115471674A (en) * | 2022-09-20 | 2022-12-13 | 浙江科达利实业有限公司 | Performance monitoring system of new energy vehicle carbon dioxide pipe based on image processing |
| CN115527134A (en) * | 2022-10-27 | 2022-12-27 | 浙江九烁光电工程技术有限公司 | Urban garden landscape lighting monitoring system and method based on big data |
| CN115526865A (en) * | 2022-09-30 | 2022-12-27 | 深圳市创瑞鑫科技有限公司 | Insulation testing method and system for heat dissipation module of notebook computer |
-
2022
- 2022-02-28 CN CN202210183221.1A patent/CN114580515A/en not_active Withdrawn
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114972327A (en) * | 2022-07-12 | 2022-08-30 | 爱尔达电气有限公司 | Semiconductor package test system and test method thereof |
| CN115471674A (en) * | 2022-09-20 | 2022-12-13 | 浙江科达利实业有限公司 | Performance monitoring system of new energy vehicle carbon dioxide pipe based on image processing |
| CN115471674B (en) * | 2022-09-20 | 2023-06-27 | 浙江科达利实业有限公司 | Performance monitoring system of new energy vehicle carbon dioxide pipe based on image processing |
| CN115526865A (en) * | 2022-09-30 | 2022-12-27 | 深圳市创瑞鑫科技有限公司 | Insulation testing method and system for heat dissipation module of notebook computer |
| CN115375691A (en) * | 2022-10-26 | 2022-11-22 | 济宁九德半导体科技有限公司 | Image-based semiconductor diffusion paper source defect detection system and method thereof |
| CN115527134A (en) * | 2022-10-27 | 2022-12-27 | 浙江九烁光电工程技术有限公司 | Urban garden landscape lighting monitoring system and method based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114580515A (en) | Neural network training method for intelligent detection of semiconductor desoldering | |
| CN111222395A (en) | Target detection method and device and electronic equipment | |
| CN113628179B (en) | PCB surface defect real-time detection method, device and readable medium | |
| CN115239642A (en) | Detection method, detection device and equipment for hardware defects in power transmission line | |
| CN114494185A (en) | Electrical equipment fault detection method based on RGB-T multi-scale feature fusion | |
| CN114863437A (en) | Text recognition method and device, electronic equipment and storage medium | |
| CN116189139A (en) | Traffic sign detection method based on Transformer | |
| CN116486393A (en) | Scene text detection method based on image segmentation | |
| CN112052730A (en) | 3D dynamic portrait recognition monitoring device and method | |
| CN117036271A (en) | Production line quality monitoring method and system thereof | |
| CN114663755A (en) | Neural network training method for intelligent detection of semiconductor wafer roughness | |
| CN115600157A (en) | Data processing method and device, storage medium and electronic equipment | |
| CN114723833A (en) | Improved YOLOV 5-based deep learning wafer solder joint detection method | |
| Yang et al. | A PCB defect detector based on coordinate feature refinement | |
| CN113139549A (en) | Parameter self-adaptive panorama segmentation method based on multitask learning | |
| CN116246287B (en) | Target object recognition method, training device and storage medium | |
| CN114708582B (en) | AI and RPA-based electric power data intelligent inspection method and device | |
| Zan et al. | Defect Identification of Power Line Insulators Based on a MobileViT‐Yolo Deep Learning Algorithm | |
| Lei et al. | Noise-robust wagon text extraction based on defect-restore generative adversarial network | |
| CN112085747B (en) | Image segmentation method based on local relation guidance | |
| CN113947102A (en) | Backbone two-path image semantic segmentation method for scene understanding of mobile robot in complex environment | |
| CN112733786A (en) | Detection method for pin connection stability of communication transformer | |
| CN112734016A (en) | Training method of neural network for detecting smoothness of surface of steel bar | |
| CN112686882A (en) | Method for detecting stability of pipeline erected on pipeline support frame | |
| CN115810020B (en) | Semantic guidance-based coarse-to-fine remote sensing image segmentation method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20220603 |