Sei thickness obtaining method, device, storage medium and program product

Abstract

The present disclosure provides an SEI thickness obtaining method, device, storage medium and program product. The SEI thickness obtaining method comprises: inputting an image containing an SEI region into a pre-trained instance segmentation model to identify the SEI region from the image by the instance segmentation model; obtaining a scale value and a scale straight line in the image to establish a correspondence between pixels and physical sizes in the image; performing curve fitting on the contour of the SEI region identified from the image to obtain a contour curve of the SEI region; and obtaining the thickness of the SEI region based on the contour curve and the correspondence between the pixels and the physical sizes. The present disclosure can improve the efficiency and accuracy of SEI region thickness analysis.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of battery detection, and in particular, to a SEI thickness obtaining method, device, storage medium and program product. BACKGROUND

[0002] The SEI (Solid Electrolyte Interface) film is an important component of a battery. In the related art, the phase evolution of the SEI film is mainly analyzed in an artificial manner. The identification of the SEI in the in-situ electron microscope image and the thickness analysis thereof depend on visual judgment, and the data analysis needs the experience of experts. However, tens of thousands of pictures are generated in one in-situ electron microscope experiment, and it is difficult to manually process them. Moreover, the artificial subjective conjecture leads to low accuracy, and the experimental results affect the subsequent research. In the related art, the efficiency and accuracy of the SEI thickness analysis in the image are low. SUMMARY

[0003] One technical problem solved by the present disclosure is that the efficiency and accuracy of the SEI thickness analysis in the image are low in the related art.

[0004] According to one aspect of the present disclosure, a SEI thickness obtaining method is provided, including: inputting an image containing a solid electrolyte interface SEI region to a pre-trained instance segmentation model to identify the SEI region from the image by the instance segmentation model; obtaining a scale value and a scale straight line in the image to establish a correspondence between pixels and physical sizes in the image; performing curve fitting on a contour of the identified SEI region to obtain a contour curve of the SEI region; and obtaining a thickness of the SEI region based on the contour curve and the correspondence between the pixels and the physical sizes.

[0005] In the above technical solution, the image is identified by the pre-trained instance segmentation model to identify the SEI region from the image, and the scale value and the scale straight line are obtained to establish the correspondence between the pixels and the physical sizes in the image. After the contour of the identified SEI region is fitted to obtain the contour curve, the thickness of the SEI region is obtained based on the contour curve and the correspondence between the pixels and the physical sizes. Compared with the artificial method, the method can improve the efficiency of the SEI region thickness analysis, and since the method does not exist the artificial subjective conjecture, the method can improve the accuracy of the SEI region thickness analysis.

[0006] In some embodiments, the profile curve includes a first curve corresponding to a first edge of the SEI region and a second curve corresponding to a second edge of the SEI region, the first edge being one edge of the SEI region extending along a long direction of the SEI region, and the second edge being an edge of the SEI region opposite to the first edge. With the first curve and the second curve in the profile curve, the thickness of the SEI region can be obtained conveniently, and the accuracy of the analysis of the thickness of the SEI region can be improved.

[0007] In some embodiments, based on the profile curve and the correspondence between the pixels and the physical sizes, the thickness of the SEI region is obtained by: obtaining, based on the first curve, a tangent line at a position point of the first edge of the SEI region; obtaining, based on the tangent line and the position point, a normal line of the first curve at the position point, wherein the normal line intersects the second curve to form an intersection point; and obtaining, based on the position point, the intersection point and the correspondence between the pixels and the physical sizes, a distance between the position point and the intersection point, wherein the distance is the thickness of the SEI region at the position point. In this embodiment, based on the first curve, the tangent line and the normal line at the position point of the first edge of the SEI region can be obtained, the intersection point of the normal line and the second curve can be obtained, and then based on the position point, the intersection point and the correspondence between the pixels and the physical sizes, the distance between the position point and the intersection point can be obtained, which can improve the accuracy of the analysis of the thickness of the SEI region.

[0008] In some embodiments, based on the position point, the intersection point and the correspondence between the pixels and the physical sizes, the distance between the position point and the intersection point is obtained by: obtaining, along the direction of the normal line, the number of pixels between the position point and the intersection point; and calculating the distance between the position point and the intersection point according to the correspondence between the pixels and the physical sizes and the number of pixels between the position point and the intersection point. In this embodiment, by obtaining the number of pixels between the position point and the intersection point, the distance between the position point and the intersection point can be calculated conveniently and accurately based on the number of pixels and the correspondence between the pixels and the physical sizes, and the accuracy of the analysis of the thickness of the SEI region can be improved.

[0009] In some embodiments, the SEI thickness obtaining method further includes: obtaining, based on a plurality of position points of the first edge of the SEI region, a plurality of thicknesses of the SEI region at the plurality of position points; and calculating an average value of the plurality of thicknesses as an average thickness of the SEI region. In this embodiment, by obtaining a plurality of thickness values of a plurality of position points, and then calculating an average value of the plurality of thicknesses as an average thickness of the SEI region, the accuracy of the analysis of the thickness of the SEI region can be further improved.

[0010] In some embodiments, obtaining the scale value and the scale straight line in the image to establish the correspondence between the pixels in the image and the physical size comprises: recognizing the scale value in the image by using a character recognition library; processing the image by using an edge detection algorithm to obtain the edge of the scale straight line in the image; processing the scale straight line in the image based on the edge of the scale straight line to obtain the number of single-row pixels corresponding to the scale straight line; and calculating the physical size corresponding to each pixel based on the scale value and the number of single-row pixels corresponding to the scale straight line to establish the correspondence between the pixels in the image and the physical size. In this way, the correspondence between the pixels and the physical size can be established, thereby facilitating subsequent accurate obtaining of the thickness of the SEI region, which can improve the accuracy of SEI region thickness analysis.

[0011] In some embodiments, the SEI thickness obtaining method further comprises: training the instance segmentation model by using images pre-labeled with the SEI region before inputting the image into the instance segmentation model. In this embodiment, training the instance segmentation model by using images pre-labeled with the SEI region can enable the instance segmentation model to identify the SEI region in the image, thereby improving the accuracy of SEI region identification.

[0012] In some embodiments, training the instance segmentation model by using images pre-labeled with the SEI region comprises: labeling the SEI region in a plurality of images by using a labeling tool to obtain a plurality of images pre-labeled with the SEI region; and inputting the plurality of images pre-labeled with the SEI region into the instance segmentation model to train the instance segmentation model. This can enable the instance segmentation model to identify the SEI region in the image, thereby improving the accuracy of SEI region identification.

[0013] In some embodiments, the instance segmentation model is a large instance segmentation model constructed by a Transformer architecture. Using a large instance segmentation model constructed by a Transformer architecture to identify the SEI region can improve the accuracy of SEI region identification.

[0014] According to another aspect of the present disclosure, there is provided an SEI thickness obtaining apparatus, comprising: an input unit configured to input an image containing a solid electrolyte interface (SEI) region to a pre-trained instance segmentation model to identify the SEI region from the image by the instance segmentation model; a first obtaining unit configured to obtain a scale value and a scale straight line in the image to establish a correspondence between pixels and physical sizes in the image; a fitting unit configured to perform curve fitting on a contour of the SEI region identified from the image to obtain a contour curve of the SEI region; and a second obtaining unit configured to obtain a thickness of the SEI region based on the contour curve and the correspondence between pixels and physical sizes. The SEI thickness obtaining apparatus can improve the efficiency and accuracy of SEI region thickness analysis.

[0015] According to another aspect of the present disclosure, there is provided an SEI thickness obtaining apparatus, comprising: a memory; and a processor coupled to the memory, the processor being configured to perform the SEI thickness obtaining method as previously described based on instructions stored in the memory.

[0016] According to another aspect of the present disclosure, there is provided a computer readable storage medium having computer instructions stored thereon, the computer instructions being executed by a processor to implement the SEI thickness obtaining method as previously described.

[0017] According to another aspect of the present disclosure, there is provided a computer program product comprising computer programs or instructions, the computer programs or instructions being executed by a processor to implement the SEI thickness obtaining method as previously described.

[0018] Other features and advantages of the present disclosure will become apparent from the following detailed description of exemplary embodiments thereof, taken in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings, which form a part of the specification, illustrate embodiments of the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0020] The present disclosure can be understood more readily by reference to the following detailed description of exemplary embodiments of the present disclosure and the attached drawings, of which:

[0021] Figure 1 is a flowchart illustrating an SEI thickness obtaining method according to some embodiments of the present disclosure;

[0022] Figure 2 is a TEM (Transmission Electron Microscope) image containing an SEI region according to some embodiments of the present disclosure;

[0023] Figure 3 is a TEM image showing a SEI region according to some embodiments of the present disclosure;

[0024] Figure 4 is a TEM image showing a SEI region according to some embodiments of the present disclosure;

[0025] Figure 5 is a partial view of a TEM image showing a SEI region according to some embodiments of the present disclosure;

[0026] Figure 6 is a schematic structural block diagram of a SEI thickness obtaining apparatus according to some embodiments of the present disclosure;

[0027] Figure 7 is a schematic structural block diagram of a SEI thickness obtaining apparatus according to some embodiments of the present disclosure;

[0028] Figure 8 is a schematic structural block diagram of a SEI thickness obtaining apparatus according to some embodiments of the present disclosure;

[0029] Figure 9 is a histogram showing a thickness distribution of a SEI region according to some embodiments of the present disclosure.

[0030] It should be understood that the dimensions of the various portions shown in the drawings are chosen for purposes of illustration only, and are not intended to limit the scope of the disclosure. In addition, like or similar designations can refer to like or similar elements. DETAILED DESCRIPTION

[0031] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. The description of the exemplary embodiments is merely illustrative in nature and is in no way intended to limit the scope of the disclosure, its application or uses. The disclosure can be implemented in numerous different forms, as will be apparent to one of ordinary skill in the art. The embodiments provided are in the nature of a best mode of the disclosure and are provided to give a full and enabling disclosure as required by the patent laws, and are not intended to limit the scope of the disclosure, its application or uses. It should be noted that the relative arrangement of the components and steps illustrated in these embodiments, the components of the compositions, the numerical expressions and numerical values set forth in these embodiments are to be interpreted as merely exemplary and are not to be construed as limiting.

[0032] The terms "first", "second", and similar terms are used herein to distinguish one element from another, and do not necessarily have an ordinal, numerical or chronological significance. The terms "comprises", "comprising", "includes", "including" and the like can be used herein to encompass the elements listed after such terms, without excluding other elements. The terms "above", "below", "left", "right" and the like are used herein merely to illustrate a relative position relationship, and can change accordingly when the absolute position of the described object changes.

[0033] In the present disclosure, when it is described that a specific device is located between a first device and a second device, there can or can not be an intervening device between the specific device and the first device or the second device. When it is described that a specific device is connected to another device, the specific device can be directly connected to the other device without an intervening device, or can not be directly connected to the other device with an intervening device.

[0034] All terms used in the present disclosure, including technical or scientific terms, have the same meanings as those understood by those skilled in the art to which the present disclosure pertains, unless otherwise specifically defined. It should also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and not in an idealized or overly formal sense, unless expressly so defined herein.

[0035] Techniques, methods, and devices known to those of ordinary skill in the relevant art can not be discussed in detail herein, but should be considered as part of the description unless otherwise noted in detail.

[0036] Figure 1 is a flowchart illustrating an SEI thickness obtaining method according to some embodiments of the present disclosure. The SEI thickness obtaining method can be performed by an SEI thickness obtaining apparatus. As shown in Figure 1 , the SEI thickness obtaining method includes steps S102 to S108.

[0037] In step S102, an image containing an SEI region is input to a pre-trained instance segmentation model to identify the SEI region from the image by the instance segmentation model.

[0038] For example, a TEM image containing an SEI region can be input to a pre-trained instance segmentation model to identify the SEI region (or referred to as SEI layer) and a non-SEI region from the image by the instance segmentation model. Here, the non-SEI region is a region in the image other than the SEI region. For example, as shown in Figure 2 or Figure 3 , the image includes an SEI region 21 and a non-SEI region 22, the SEI region 21 includes a first edge 211 and a second edge 212 extending in the long strip direction of the extending SEI region. In addition, as shown in Figure 2 or Figure 3 , a ruler value 32 and a ruler straight line 31 are also included in the image.

[0039] Here, the instance segmentation model is a machine learning model known to those skilled in the art. For example, the instance segmentation model is an instance segmentation large model constructed by a Transformer architecture. Using the instance segmentation large model constructed by the Transformer architecture to identify the SEI region can improve the accuracy of SEI region identification.

[0040] In step S104, the scale value and the scale straight line in the image are obtained to establish the correspondence between the pixels in the image and the physical size.

[0041] In some embodiments, this step S104 can include identifying the scale value in the image using a text recognition library. For example, as shown in FIG. 3B, the scale value 32 (e.g., 20 nm (nanometer)) can be identified using the text recognition library EasyOCR. Figure 4

[0042] This step S104 can also include processing the image using an edge detection algorithm to obtain the edge of the scale straight line in the image. For example, as shown in FIG. 3B, the Canny edge detection algorithm is used to process the image to obtain the edge of the scale straight line 31 in the image, which is composed of pixels. Figure 4

[0043] This step S104 can also include processing the scale straight line in the image based on the edge of the scale straight line to obtain the number of single-row pixels corresponding to the scale straight line. For example, processing the scale straight line in the image using the Hough transform detection method can obtain the pixels in the region where the scale straight line is located, which can include one or more rows of pixels. From the one or more rows of pixels, a single row of pixels (i.e., one row of pixels) is obtained, and the number of single-row pixels is further obtained.

[0044] This step S104 can also include calculating the physical size corresponding to each pixel based on the scale value and the number of single-row pixels corresponding to the scale straight line to establish the correspondence between the pixels in the image and the physical size. Here, the ratio of the scale value to the number of single-row pixels is calculated, which is the physical size corresponding to each pixel, thereby establishing the correspondence between the pixels and the physical size.

[0045] In this way, by obtaining the scale value and the scale straight line in the image, the correspondence between the pixels in the image and the physical size can be established. This can facilitate subsequent accurate acquisition of the thickness of the SEI region, and thus can improve the accuracy of SEI region thickness analysis.

[0046] ​​In the above method, the scale value is recognized by a character recognition library, and the scale straight line is detected by an edge detection algorithm and Hough transform, and the length mark (for example, "20 nm" or "200 nm", etc.) can be recognized. The conversion relationship between the pixel length of the scale and the actual physical size is automatically calculated, and all input images are scaled to a unified physical size (for example, 1 pixel = 0.5 nm) according to the scale information. This method can output a unified resolution, reduce the size difference caused by different devices and different magnifications, realize the normalization of image size, and improve the comparability of measurement results.

[0047] In the related art, when performing SEI thickness analysis, different magnifications (such as ×100k, ×300k, ×500k, etc.) are used for collecting electron microscope images, resulting in inconsistent scales of the images, and it is not possible to directly perform unified SEI thickness recognition and measurement on images with different magnifications.

[0048] In the method of the embodiments of the present disclosure, the scale text and scale are accurately positioned and recognized by using character recognition, edge detection and scale detection algorithms, and the scale information can be accurately extracted under different imaging qualities. By establishing the correspondence between the pixels and the physical size (as a pixel-physical size conversion model), the TEM images obtained by different devices and different magnifications are unified to a standard size coordinate system, so as to unify the image size as much as possible.

[0049] In step S106, the contour of the SEI region recognized from the image is curve-fitted to obtain a contour curve of the SEI region.

[0050] For example, the contour of the SEI region can be curve-fitted by using a spline function to obtain the contour curve of the SEI region. For example, as shown in Figure 5 The contour curve includes a first curve 41 corresponding to the first edge of the SEI region and a second curve 42 corresponding to the second edge of the SEI region. As an example, Figure 5 Only a part of the first curve 41 and a part of the second curve 42 are shown in the above. As described before, the first edge (for example, the first edge 211) is one edge of the SEI region extending along the strip direction of the SEI region, and the second edge (for example, the second edge 212) is the edge opposite to the first edge of the SEI region. By using the first curve and the second curve in the contour curve, the thickness of the SEI region can be conveniently obtained, and the accuracy of the SEI region thickness analysis can be improved.

[0051] In step S108, the thickness of the SEI region is obtained based on the contour curve and the correspondence between the pixels and the physical size.

[0052] In some embodiments, step S108 can include: obtaining a tangent line at a position point of the first edge of the SEI region based on the first curve; obtaining a normal line of the first curve at the position point based on the tangent line and the position point, wherein the normal line intersects the second curve to form an intersection point; and obtaining a distance between the position point and the intersection point based on the position point, the intersection point, and the correspondence between the pixels and the physical dimensions, wherein the distance is a thickness of the SEI region at the position point.

[0053] For example, as shown in FIG. 4, a tangent line 51 at a position point 411 of the first edge of the SEI region is obtained based on the first curve 41. Here, the position point 411 is also a point on the first curve 41. Then, a normal line 61 of the first curve 41 at the position point 411 can be obtained based on the tangent line 51 and the position point 411, wherein the normal line 61 intersects the second curve 42 to form an intersection point 421. In addition, as shown in FIG. 4, a tangent line 52 of the second curve 42 at the intersection point 421 is also shown. Figure 5 Figure 5 Next, as shown in FIG. 4, a distance d between the position point 411 and the intersection point 421 is obtained based on the position point 411, the intersection point 421, and the correspondence between the pixels and the physical dimensions, wherein the distance d is a thickness of the SEI region at the position point. Here, the thickness of the SEI region at the position point is a dimension of the SEI region at the position point and extending along the direction of the normal line. In other words, the SEI thickness is a vertical distance between the two edges of the SEI layer along the long direction. Figure 5

[0054] In this embodiment, the tangent line and the normal line at a position point of the first edge of the SEI region are obtained based on the first curve, the intersection point of the normal line and the second curve is obtained, and then the distance between the position point and the intersection point is obtained based on the position point, the intersection point, and the correspondence between the pixels and the physical dimensions, which can improve the accuracy of the SEI region thickness analysis.

[0055] In some embodiments, obtaining the distance between the position point and the intersection point based on the position point, the intersection point, and the correspondence between the pixels and the physical dimensions includes: obtaining a number of pixels between the position point and the intersection point along the direction of the normal line; and calculating the distance between the position point and the intersection point according to the correspondence between the pixels and the physical dimensions and the number of pixels between the position point and the intersection point. That is, after the profile of the SEI region is curve-fitted to establish the reference tangent line equation, pixel-level scanning is performed along the direction of the normal line of the SEI region profile tangent line, so as to obtain the vertical distance between the two edges (or boundaries) of the SEI layer as the thickness of the SEI region.

[0056] ​​In the above embodiment, by obtaining the number of pixels between the position point and the intersection point, the distance between the position point and the intersection point can be conveniently and accurately calculated based on the number of pixels and the correspondence between the pixels and the physical size (i.e., the size of each pixel), thereby improving the accuracy of the thickness analysis of the SEI region.

[0057] In some embodiments, a plurality of position points can be determined in the vicinity of the above position point, and a corresponding tangent line and normal line can be obtained at each position point, so as to obtain a plurality of normal lines. The average value of the slopes of the plurality of normal lines is calculated as an average slope, and the average value of the intercepts of the plurality of normal lines is calculated as an average intercept. A new normal line is determined by the average slope and the average intercept, which is the average normal line of the plurality of normal lines. The thickness of the SEI region determined along the direction of the average normal line is the thickness at the above position point and the positions in the vicinity thereof, which is more accurate.

[0058] So far, the SEI thickness obtaining method according to some embodiments of the present disclosure is provided. The method comprises: inputting an image containing a solid electrolyte interface (SEI) region into a pre-trained instance segmentation model to identify the SEI region from the image by the instance segmentation model; obtaining a scale value and a scale straight line in the image to establish a correspondence between pixels in the image and physical sizes; performing curve fitting on the contour of the SEI region identified from the image to obtain a contour curve of the SEI region; and obtaining the thickness of the SEI region based on the contour curve and the correspondence between the pixels and the physical sizes. In the method, the image is identified by the pre-trained instance segmentation model to identify the SEI region from the image, and the scale value and the scale straight line are obtained to establish the correspondence between the pixels in the image and the physical sizes. After the contour curve of the identified SEI region is obtained by fitting the contour, the thickness of the SEI region is obtained based on the contour curve and the correspondence between the pixels and the physical sizes. Compared with the manual method, the method can improve the efficiency of the SEI region thickness analysis, and since the method does not involve artificial subjective speculation, the method can improve the accuracy of the SEI region thickness analysis. For example, the method can accurately measure the thickness of a curved SEI.

[0059] In the embodiments of the present disclosure, by the accurate segmentation of the SEI region in the TEM image and the automatic identification algorithm of the scale, the size normalization of the image is realized, and the thickness parameters of the SEI are quantified. Compared with the manual analysis, the method of the present disclosure can shorten the processing time of a single image, for example, the processing time of a single image is shortened from 10 minutes to 1 second, and batch processing is supported, which accelerates the research on the dynamic evolution process of the SEI, realizes the establishment of the SEI model, and determines the evolution mechanism.

[0060] In some cases, the above method can solve the cross-section SEI measurement problem to a certain extent, and realize full-area automatic pixel-level scanning. For example, a fully automated measurement process can realize more than 1000 measurement points per micrometer of SEI length, and the data volume is improved by 2 to 3 orders of magnitude compared with the manual single-point measurement method. In some embodiments, the thickness distribution atlas generated by the system automatically includes multiple parameters: average thickness (for example, accurate to 0.1 nm), thickness variance (reflecting uniformity), maximum / minimum value (identifying abnormal areas), and gradient distribution, etc. Key parameters provide comprehensive quantitative basis for SEI performance evaluation.

[0061] In some embodiments, the SEI thickness obtaining method can further include: obtaining a plurality of thicknesses of the SEI region at a plurality of position points on the first edge of the SEI region based on the plurality of position points; and calculating an average value of the plurality of thicknesses as the average thickness of the SEI region. That is, a plurality of thicknesses of the SEI region can be obtained by selecting a plurality of position points to calculate the corresponding thicknesses respectively, and the average thickness of the SEI region can be calculated by calculating the average value of the plurality of thicknesses, which can further improve the accuracy of SEI region thickness analysis.

[0062] Here, the system can automatically perform a large number of normal line measurements. For example, 1000 measurement points (i.e. the position points described above) can be detected (for example, the length between two adjacent measurement points is 1 nm), so that 1000 thickness data are obtained, and statistical results such as thickness distribution histogram can be output by excluding abnormal value interference through statistical methods. Through full-area pixel-level measurement and data statistics, a thickness distribution feature atlas is established.

[0063] For example, Figure 9 The thickness distribution histograms of two SEIs (SEI1 and SEI2) are shown, so that the thickness conditions of the two SEIs can be compared, and the SEIs can be analyzed accordingly.

[0064] In some embodiments, the SEI thickness obtaining method can further include: training the instance segmentation model using images pre-labeled with SEI regions before inputting the image to the instance segmentation model. In this embodiment, training the instance segmentation model using images pre-labeled with SEI regions can enable the instance segmentation model to identify the SEI region in the image, improving the accuracy of SEI region identification.

[0065] In some embodiments, training the instance segmentation model using images pre-labeled with SEI regions includes: labeling the SEI regions in a plurality of images using a labeling tool to obtain a plurality of images pre-labeled with SEI regions; and inputting the plurality of images pre-labeled with SEI regions into the instance segmentation model to train the instance segmentation model.

[0066] For example, performing high-resolution transmission electron microscopy data acquisition and labeling, a high-resolution TEM image can be obtained using a transmission electron microscope, and the SEI region is labeled using the labelme software to accurately define the SEI / electrode interface; provide SEI labeling data to the machine learning model for training to distinguish SEI regions and non-SEI regions.

[0067] For example, only key frames (for example, randomly select one image out of every 100 images) can be labeled, and the SEI region in the image can be labeled through a semi-automatic labeling tool to facilitate subsequent training operations. For example, as shown in Figure 2 The red box is the manually labeled SEI region. In this way, an automated recognition system is established to achieve intelligent recognition of SEI. A large instance segmentation model constructed by introducing a transformer architecture is used to recognize SEI regions using the good feature extraction capability and generalization of the large instance segmentation model. Provide SEI labeling data to the machine learning model for training to distinguish SEI regions and non-SEI regions. Figure 3 The green area in the above figure is the SEI region labeled by the trained instance segmentation model.

[0068] Generally, in-situ liquid electrochemical experiments are difficult, and high-quality in-situ electron microscopy data are relatively scarce. In the above training method, a small amount of labeled data (for example, about 50 to 100 images) can be used to fine-tune the instance segmentation model of the SEI region to obtain a high-precision SEI region instance segmentation model.

[0069] Through the above training process, an instance segmentation model for efficient and intelligent SEI recognition can be obtained. Through the key frame labeling strategy, the labeling quality is improved while reducing the manual labeling workload by more than 90%, achieving high-precision SEI recognition under small sample conditions, and the segmentation accuracy can reach more than 95%.

[0070] Figure 6 is a structural block diagram schematically showing an SEI thickness obtaining apparatus according to some embodiments of the present disclosure. As Figure 6 shown, the SEI thickness obtaining apparatus includes an input unit 602, a first obtaining unit 604, a fitting unit 606, and a second obtaining unit 608.

[0071] The input unit 602 is configured to input an image containing a solid electrolyte interface (SEI) region to a pre-trained instance segmentation model to identify the SEI region from the image through the instance segmentation model.

[0072] The first obtaining unit 604 is configured to obtain a scale value and a scale straight line in the image, so as to establish a correspondence between pixels in the image and physical sizes.

[0073] The fitting unit 606 is configured to perform curve fitting on the contour of the SEI region identified from the image, so as to obtain a contour curve of the SEI region.

[0074] The second obtaining unit 608 is configured to obtain a thickness of the SEI region based on the contour curve and the correspondence between the pixels and the physical sizes.

[0075] In this way, the SEI thickness obtaining apparatus can improve the efficiency and accuracy of SEI region thickness analysis.

[0076] In some embodiments, the contour curve includes a first curve corresponding to a first edge of the SEI region and a second curve corresponding to a second edge of the SEI region, the first edge being one edge of the SEI region extending along a strip direction of the SEI region, and the second edge being an edge of the SEI region opposite to the first edge.

[0077] In some embodiments, the second obtaining unit 608 is configured to: based on the first curve, obtain a tangent line at a position point of the first edge of the SEI region; based on the tangent line and the position point, obtain a normal line of the first curve at the position point, wherein the normal line intersects with the second curve to form an intersection point; and based on the position point, the intersection point, and the correspondence between the pixels and the physical sizes, obtain a distance between the position point and the intersection point, wherein the distance is a thickness of the SEI region at the position point.

[0078] In some embodiments, the second obtaining unit 608 is configured to: along a direction of the normal line, obtain a number of pixels between the position point and the intersection point; and according to the correspondence between the pixels and the physical sizes and the number of pixels between the position point and the intersection point, calculate the distance between the position point and the intersection point.

[0079] In some embodiments, the second obtaining unit 608 is further configured to: based on a plurality of position points of the first edge of the SEI region, obtain a plurality of thicknesses of the SEI region at the plurality of position points; and calculate an average value of the plurality of thicknesses as an average thickness of the SEI region.

[0080] In some embodiments, the first obtaining unit 604 is configured to: identify the scale value in the image by using a character recognition library; process the image by using an edge detection algorithm to obtain edges of the scale straight line in the image; process the scale straight line in the image based on the edges of the scale straight line, to obtain a number of single-row pixels corresponding to the scale straight line; and based on the scale value and the number of single-row pixels corresponding to the scale straight line, calculate a physical size corresponding to each pixel, to establish the correspondence between the pixels in the image and the physical sizes.

[0081] In some embodiments, the SEI thickness obtaining apparatus further comprises a training unit configured to train the instance segmentation model by using images of pre-labeled SEI regions.

[0082] In some embodiments, the training unit is configured to label SEI regions in the plurality of images by using a labeling tool to obtain the plurality of images of pre-labeled SEI regions, and input the plurality of images of pre-labeled SEI regions into the instance segmentation model to train the instance segmentation model.

[0083] In some embodiments, the instance segmentation model is a large instance segmentation model constructed by a Transformer architecture.

[0084] In the above method and apparatus of the embodiments of the present disclosure, the SEI region is accurately segmented by the intelligent image processing method, the image size normalization is realized by the ruler recognition, the thickness distribution feature map is established by the full-region pixel-level measurement and data statistics, and a quantitative analysis tool is provided for the battery material interface research. The artificial intelligence and electron microscope image processing technology are combined, the machine learning is developed for data analysis, and a precise and efficient data analysis tool is provided, which can accelerate the research on the dynamic evolution process of SEI and realize the establishment of SEI model and determination of evolution mechanism.

[0085] Figure 7 is a structural block diagram schematically showing an SEI thickness obtaining apparatus according to some other embodiments of the present disclosure. The SEI thickness obtaining apparatus comprises a memory 710 and a processor 720. Wherein:

[0086] The memory 710 can be a disk, a flash memory or any other non-volatile storage medium. The memory is used to store Figure 1 instructions in the corresponding embodiments.

[0087] The processor 720 is coupled to the memory 710 and can be implemented as one or more integrated circuits, such as a microprocessor or a microcontroller. The processor 720 is used to execute the instructions stored in the memory, which can improve the efficiency and accuracy of SEI region thickness analysis.

[0088] In one embodiment, as Figure 8 shown, the SEI thickness obtaining apparatus 700 comprises a memory 710 and a processor 720. The processor 720 is coupled to the memory 710 through a BUS 730. The SEI thickness obtaining apparatus 700 can also be connected to an external storage device 750 through a storage interface 740 to call external data, and can also be connected to a network or another computer system (not shown) through a network interface 760, which will not be described in detail here.

[0089] In the embodiment, the data instruction is stored in the memory, and the processor processes the data instruction, so that the efficiency and accuracy of the SEI area thickness analysis are improved.

[0090] In another embodiment, the disclosure also provides a computer readable storage medium (for example, a non-transitory computer readable storage medium) having stored thereon computer program instructions, which, when executed by a processor, implement the steps of the method in the corresponding embodiment. Figure 1 Those skilled in the art should understand that the embodiments of the disclosure can be provided as a method, device or computer program product. Therefore, the disclosure can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the disclosure can take the form of a computer program product implemented on one or more computer-usable non-transitory storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0091] The disclosure is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the disclosure. It should be understood that each flow and / or block in the flowcharts and / or block diagrams and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. Figure 1 The device that implements the functions specified in one or more flows and / or blocks.

[0092] These computer program instructions can also be stored in a computer readable memory that can guide the computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer readable memory produce a product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. Figure 1 The device that implements the functions specified in one or more flows and / or blocks.

[0093] These computer program instructions can also be loaded into a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. ​ The device that implements the functions specified in one or more flows and / or blocks.

[0094] In some embodiments of the present disclosure, a computer program product is also provided, which comprises a computer program or instructions, which, when executed by a processor, implement the SEI thickness obtaining method as described above.

[0095] In some embodiments of the present disclosure, a computer program is also provided, which comprises instructions, which, when executed by a processor, cause the processor to perform the SEI thickness obtaining method as described above.

[0096] So far, the embodiments of the present disclosure have been described in detail. In order to avoid obscuring the concept of the present disclosure, some details known in the art are not described. Those skilled in the art can fully understand how to implement the technical solutions disclosed herein according to the above description.

[0097] Although some specific embodiments of the present disclosure have been described in detail through examples, those skilled in the art should understand that the above examples are only for illustration, not for limiting the scope of the present disclosure. Those skilled in the art should understand that the above embodiments can be modified or some technical features can be replaced equivalently without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A method for obtaining SEI thickness, characterized in that, include: An image containing a solid electrolyte interface (SEI) region is input into a pre-trained instance segmentation model to identify the SEI region from the image using the instance segmentation model, wherein the instance segmentation model is a large instance segmentation model built using a Transformer architecture. Obtain the scale values ​​and scale lines in the image to establish the correspondence between pixels in the image and physical dimensions; Curve fitting is performed on the contour of the SEI region identified from the image to obtain the contour curve of the SEI region; and The thickness of the SEI region is obtained based on the contour curve and the correspondence between the pixels and physical dimensions; The contour curve includes a first curve corresponding to a first edge of the SEI region and a second curve corresponding to a second edge of the SEI region. The first edge is an edge of the SEI region extending along the elongated direction of the SEI region, and the second edge is an edge of the SEI region opposite to the first edge. The thickness of the SEI region is obtained based on the contour curve and the correspondence between pixels and physical dimensions, including: obtaining a tangent at a location point on a first edge of the SEI region based on the first curve; obtaining a normal to the first curve at the location point based on the tangent and the location point, wherein the normal intersects the second curve to form an intersection point; obtaining the number of pixels between the location point and the intersection point along the direction of the normal; and calculating the distance between the location point and the intersection point according to the correspondence between pixels and physical dimensions and the number of pixels between the location point and the intersection point, wherein the distance is the thickness of the SEI region at the location point; Obtaining the scale values ​​and scale lines in the image to establish the correspondence between pixels and physical dimensions in the image includes: recognizing the scale values ​​in the image using a character recognition library; processing the image using an edge detection algorithm to obtain the edges of the scale lines in the image; processing the scale lines in the image based on the edges of the scale lines to obtain the number of pixels in a single row corresponding to the scale lines; and calculating the physical dimension corresponding to each pixel based on the scale values ​​and the number of pixels in a single row corresponding to the scale lines, to establish the correspondence between pixels and physical dimensions in the image.

2. The method for obtaining SEI thickness according to claim 1, characterized in that, Also includes: Based on multiple location points at the first edge of the SEI region, multiple thicknesses of the SEI region at the multiple location points are obtained; and Calculate the average value of the multiple thicknesses as the average thickness of the SEI region.

3. The method for obtaining SEI thickness according to claim 1, characterized in that, Also includes: Before inputting the image into the instance segmentation model, the instance segmentation model is trained using an image with the SEI region pre-labeled.

4. The method for obtaining SEI thickness according to claim 3, characterized in that, Training the instance segmentation model using images pre-annotated with the SEI region includes: The SEI regions in multiple images are labeled using a labeling tool to obtain multiple images with pre-labeled SEI regions; and Multiple images with pre-labeled SEI regions are input into the instance segmentation model to train the instance segmentation model.

5. An SEI thickness obtaining device, characterized in that, include: An input unit is used to input an image containing a solid electrolyte interface (SEI) region into a pre-trained instance segmentation model to identify the SEI region from the image through the instance segmentation model, wherein the instance segmentation model is a large instance segmentation model constructed using a Transformer architecture. The first obtaining unit is used to obtain the scale values ​​and scale lines in the image, so as to establish the correspondence between pixels in the image and physical dimensions; A fitting unit is configured to perform curve fitting on the contour of the SEI region identified from the image to obtain the contour curve of the SEI region; and The second obtaining unit is used to obtain the thickness of the SEI region based on the contour curve and the correspondence between the pixels and physical dimensions; The contour curve includes a first curve corresponding to a first edge of the SEI region and a second curve corresponding to a second edge of the SEI region. The first edge is an edge of the SEI region extending along the elongated direction of the SEI region, and the second edge is an edge of the SEI region opposite to the first edge. The second obtaining unit is configured to: obtain a tangent line at a location point on a first edge of the SEI region based on the first curve; obtain a normal line of the first curve at the location point based on the tangent line and the location point, wherein the normal line intersects the second curve to form an intersection point; and obtain a distance between the location point and the intersection point based on the location point, the intersection point, and the correspondence between the pixel and the physical size, wherein the distance is the thickness of the SEI region at the location point; The second obtaining unit is configured to: obtain the number of pixels between the location point and the intersection point along the direction of the normal; and calculate the distance between the location point and the intersection point based on the correspondence between the pixels and the physical size and the number of pixels between the location point and the intersection point; The first obtaining unit is configured to: identify the scale value in the image using a character recognition library; process the image using an edge detection algorithm to obtain the edge of the scale line in the image; process the scale line in the image based on the edge of the scale line to obtain the number of pixels in a single row corresponding to the scale line; and calculate the physical size corresponding to each pixel based on the scale value and the number of pixels in a single row corresponding to the scale line, so as to establish the correspondence between pixels and physical sizes in the image.

6. An SEI thickness obtaining device, characterized in that, include: Memory; as well as A processor coupled to the memory, the processor being configured to execute the SEI thickness acquisition method as described in any one of claims 1 to 4 based on instructions stored in the memory.

7. A computer-readable storage medium, characterized in that, It stores computer instructions that, when executed by a processor, implement the SEI thickness acquisition method as described in any one of claims 1 to 4.

8. A computer program product, characterized in that, The computer program product includes a computer program or instructions that, when executed by a processor, implement the SEI thickness acquisition method as described in any one of claims 1 to 4.

Images