Image scanning device and scanning method based on image recognition model training
By designing an image scanning device that can adjust the inclination angle and position, the problem of unclear and incomplete scanning of larger volume objects in the prior art is solved, and high-precision image scanning and model training effects are achieved.
Patent Information
- Application Number
- CN202510137328.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-07
- Publication Date
- 2025-05-09
AI Technical Summary
When scanning larger volume objects, the prior art needs to increase the distance between the object and the scanner, which makes it impossible for the scanner to scan images clearly and comprehensively, affecting the model training effect, and moving the scanners in large areas is prone to deviations and reducing accuracy.
An image scanning device is designed, including a bearing unit and a scanning unit. The load bearing unit can rotate and reciprocate in a horizontal plane to adjust the orientation and position of the object; the scanning unit consists of a first scanner and two second scanners, the first scanner can adjust the inclination angle, the second scanner can be close to each other or away from the back, and adjust the inclination angle to achieve a comprehensive scanning of the front and sides of the object.
By adjusting the inclination and position of the scanner in real time, the images of larger volumes can be clearly and comprehensively scanned, scanning accuracy can be improved, deviations caused by large-scale movements can be avoided, the motion trajectory of the scanner can be optimized, and the accuracy and efficiency of model training can be improved.
Smart Images

Figure CN119967101A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image recognition model training, and in particular to an image scanning device and a scanning method based on image recognition model training. Background Art
[0002] Image recognition model training refers to training the model using a large amount of labeled data so that it can automatically identify and classify objects or scenes in images. Model training is the core step in image recognition model training. During the training process, the model will learn based on the input image data and corresponding labels, and adjust internal parameters to minimize the loss function. Various optimization algorithms can be used during the training process.
[0003] At present, when training an image recognition model, an object is scanned by a scanning unit to obtain image features, and the obtained image features are used as training samples and passed to the image recognition model for model training, thereby completing the training of the image recognition model. When existing image scanning devices (such as an image scanning device for graphic recognition disclosed in application No. 202410442018.0) perform image scanning on objects of larger volumes, due to the limited scanning range, it is necessary to increase the distance between the object and the scanner. However, after the distance between the object and the scanner is increased, it is easy for the scanner to fail to scan the image clearly, affecting the effect of subsequent model training. Moreover, only increasing the distance between the object and the scanner will also cause the scanner to fail to scan the image comprehensively, increasing the workload. In addition, in the process of moving the scanner over a large range, the scanner is prone to deviation, affecting the scanning effect of the scanner on the image and reducing the accuracy of subsequent model training. Summary of the invention
[0004] The purpose of the present invention is to overcome the above-mentioned technical deficiencies and to propose an image scanning device and a scanning method based on image recognition model training, so as to solve the problem in the prior art that when scanning an object of a larger volume, it is necessary to increase the distance between the object and the scanner. This method may easily cause the scanner to be unable to scan the image clearly and comprehensively, affecting the effect of subsequent model training. In addition, in the process of moving the scanner over a large range, the scanner is prone to deviations, affecting the scanning effect of the scanner on the image and reducing the accuracy of subsequent model training. Technical problems.
[0005] In order to achieve the above technical objectives, the technical solution of the present invention provides an image scanning device based on image recognition model training, comprising: A carrying unit, on which the object to be scanned is placed, the carrying unit can rotate in a horizontal plane to adjust the orientation of each side of the object to be scanned, and the carrying unit can also reciprocate in a linear manner in the horizontal plane to adjust the position of the object to be scanned; The scanning unit includes a first scanner and two second scanners, wherein the first scanner is arranged at one end of the reciprocating linear motion track of the carrying unit, and is used to scan the front side of the object to be scanned, and the inclination angle of the first scanner is adjustable; the two second scanners are relatively arranged on both sides of the reciprocating linear motion track of the carrying unit and are close to the first scanner, and are respectively used to scan the two side surfaces of the object to be scanned, the two second scanners can be close to each other or away from each other, and the inclination angles of the two second scanners are adjustable.
[0006] Furthermore, the carrying unit includes a carrier, a rotation drive component and a translation drive component. The carrier is used to place the object to be scanned, the rotation drive component is connected to the carrier and is used to drive the carrier to rotate in a horizontal plane, and the translation drive component is connected to the rotation drive component and is used to drive the rotation drive component to move reciprocatingly in a horizontal plane.
[0007] Furthermore, the rotation drive assembly includes a connecting seat, a sliding seat, a connecting frame and a first rotation drive member, the bottom of the sliding seat is fixedly connected to the top of the connecting seat, the connecting frame is arranged above the sliding seat and fixedly connected to the bottom of the carrier, the first rotation drive member is arranged between the sliding seat and the connecting frame, the first rotation drive member is fixedly connected to the top of the sliding seat, and the output shaft of the first rotation drive member is fixedly connected to the connecting frame for driving the connecting frame to rotate in a horizontal plane.
[0008] Furthermore, the translation drive assembly includes a base plate, a screw rod and a second rotation drive member, a receiving groove is provided on the base plate, the receiving groove is a long strip structure, the screw rod is arranged in the receiving groove along the length direction of the receiving groove, both ends of the screw rod are rotatably connected to the base plate, the output shaft of the second rotation drive member is coaxially fixedly connected to one end of the screw rod, and is used to drive the screw rod to rotate forward or reversely, the connecting seat is slidably arranged in the receiving groove, a screw hole is provided on the connecting seat, and the connecting seat is sleeved on the screw rod through the screw hole, the screw hole is threadedly connected to the screw rod, and the sliding seat is arranged outside the receiving groove and is slidably connected to the top of the base plate.
[0009] Furthermore, the image scanning device based on image recognition model training also includes a recognition unit, which is electrically connected to the first scanner and the two second scanners to obtain image features scanned by the first scanner and the two second scanners, and use the image features as training samples for model training.
[0010] Furthermore, the scanning unit also includes a mounting frame, which is arranged at one end of the reciprocating linear movement track of the carrying unit, and the first scanner is fixedly connected to the top of the mounting frame.
[0011] Furthermore, the scanning unit also includes a spacing adjustment component, which is disposed on the mounting frame and connected to both of the two second scanners, and is used to drive the two second scanners to retract toward each other or to open away from each other.
[0012] Furthermore, the spacing adjustment component includes two swing arms, a bidirectional telescopic drive member and two transmission members, the two swing arms are relatively arranged on both sides of the mounting frame, one end of the two swing arms are hinged to the mounting frame, the two second scanners are fixedly connected to the other ends of the two swing arms in a one-to-one manner, the bidirectional telescopic drive member is horizontally arranged between the two swing arms in a direction perpendicular to the reciprocating linear movement trajectory of the bearing unit, and is fixedly connected to the mounting frame, one end of the two transmission members are connected to the two telescopic axes of the bidirectional telescopic drive member in a one-to-one manner, and the other end of the two transmission members are hinged to the middle parts of the two swing arms in a one-to-one manner, so as to convert the synchronous telescopic movement of the two telescopic axes of the bidirectional telescopic drive member into the mutual retraction or back-to-back opening of the two swing arms.
[0013] Furthermore, the scanning unit also includes an inclination adjustment component, which is connected to the bottom of the mounting frame and is used to drive the mounting frame to reciprocate in a vertical plane around a rotating axis at the bottom thereof to adjust the inclination of the mounting frame.
[0014] The present invention also provides an image scanning method based on image recognition model training, which is applicable to the image scanning device based on image recognition model training, and comprises the following steps: Classify the objects to be scanned into large objects and small objects according to preset standards; When scanning the large object, the large object is placed on the carrying unit, the two second scanners are moved away from each other to expand the scanning range, the carrying unit moves linearly in a horizontal plane to drive the large object to move to the scanning unit, the carrying unit rotates in a horizontal plane to adjust the orientation of each side of the large object, the first scanner and the two second scanners are started at the same time, the first scanner scans the front of the large object, and the two second scanners scan the two sides of the large object respectively, while the first scanner is scanning the front of the large object, the inclination angle of the first scanner is adjusted in real time to fully scan the front of the large object, and while the second scanner is scanning the side of the large object, the inclination angle of the second scanner is adjusted in real time to fully scan the side of the large object; When scanning the small object, the small object is placed on the carrying unit, the two second scanners are moved close to each other to narrow the scanning range, the carrying unit moves linearly in a horizontal plane to drive the small object to move to the scanning unit, the carrying unit rotates in a horizontal plane to adjust the orientation of each side of the small object, the first scanner and the two second scanners are started at the same time, the first scanner scans the front of the small object, and the two second scanners scan the two sides of the small object respectively, while the first scanner is scanning the front of the small object, the inclination angle of the first scanner is adjusted in real time to fully scan the front of the small object, and while the second scanner is scanning the side of the small object, the inclination angle of the second scanner is adjusted in real time to fully scan the side of the small object.
[0015] Compared with the prior art, the beneficial effects of the present invention include: when the first scanner is scanning the front side of the object to be scanned, the inclination angle of the first scanner is adjusted in real time, so that the front side of the object to be scanned can be fully scanned; when the second scanner is scanning the side of the object to be scanned, the inclination angle of the second scanner is adjusted in real time, so that the side of the object to be scanned can be fully scanned, thereby expanding the scanning range of the object to be scanned, improving the scanning accuracy of the scanner for the image features of the object to be scanned, enabling the scanner to scan the image clearly, ensuring the effect of subsequent model training, and avoiding large-scale movement of the scanner during the entire scanning process, optimizing the movement trajectory of the scanner, ensuring the scanning effect of the scanner on the image, and facilitating improving the accuracy of subsequent model training; during the scanning process, the carrying unit rotates in a horizontal plane, and the orientation of each side of the object to be scanned can be adjusted, so that the image features of the object can be acquired in all directions, which is conducive to improving the training efficiency of subsequent model training. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It is a three-dimensional structural schematic diagram of an image scanning device based on image recognition model training provided by the present invention; Figure 2 It is a three-dimensional structural schematic diagram of a carrying unit of an image scanning device based on image recognition model training provided by the present invention; Figure 3 It is a schematic diagram of the three-dimensional structure of a carrier unit of an image scanning device based on image recognition model training provided by the present invention after omitting the translation drive component; Figure 4 It is a schematic diagram of the three-dimensional structure of an image scanning device based on image recognition model training provided by the present invention after omitting the carrying unit; Figure 5 It is a schematic diagram of the three-dimensional structure of an image scanning device based on image recognition model training provided by the present invention in another viewing angle after omitting the carrying unit; Figure 6 It is a three-dimensional structural schematic diagram of the connection relationship between a spacing adjustment component of an image scanning device based on image recognition model training and a second scanner provided by the present invention; In the figure: 100-carrying unit, 110-carrying platform, 120-rotation driving component, 121-connecting seat, 122-sliding seat, 123-connecting frame, 124-first rotation driving member, 130-translation driving component, 131-bottom plate, 1311-accommodating groove, 132-screw rod, 133-second rotation driving member, 200-scanning unit, 210-first scanner, 220-second scanner, 230-mounting frame, 231-slide groove, 240-spacing adjustment component, 241-swing arm, 242-bidirectional telescopic driving member, 243-transmission member, 2431-sliding block, 2432-first connecting rod, 2433-second connecting rod, 250-tilt adjustment component, 251-base, 252-rotation axis, 253-third rotation driving member, 300-identification unit. DETAILED DESCRIPTION
[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0018] The present invention provides an image scanning device based on image recognition model training, the structure of which is as follows: Figure 1 - Figure 5As shown, it includes a carrying unit 100 and a scanning unit 200. The carrying unit 100 is used to place the object to be scanned. The carrying unit 100 can rotate in a horizontal plane to adjust the orientation of each side of the object to be scanned. The carrying unit 100 can also move back and forth in a horizontal plane to adjust the position of the object to be scanned; the scanning unit 200 includes a first scanner 210 and two second scanners 220. The first scanner 210 is arranged at one end of the reciprocating linear movement track of the carrying unit 100 to scan the front side of the object to be scanned. The inclination angle of the first scanner 210 is adjustable. The two second scanners 220 are relatively arranged on both sides of the reciprocating linear movement track of the carrying unit 100 and are close to the first scanner 210. They are used to scan the two sides of the object to be scanned respectively. The two second scanners 220 can approach each other or face away from each other, and the inclination angles of the two second scanners 220 can be adjusted.
[0019] When in use, according to the size of the object, the two second scanners 220 are moved closer to each other or away from each other, and the distance between the two second scanners 220 is adjusted to a suitable value. Then, the carrying unit 100 moves linearly in a horizontal plane, which can drive the object to be scanned to the scanning unit 200. Then, the carrying unit 100 rotates in a horizontal plane, which can adjust the direction of each side of the object to be scanned. At the same time, the first scanner 210 and the two second scanners 220 are started. The first scanner 210 scans the front of the object to be scanned, and the two second scanners 220 scan the two sides of the object to be scanned respectively. When the first scanner 210 scans the front of the object to be scanned, the inclination angle of the first scanner 210 is adjusted in real time, so that the object to be scanned can be adjusted. The front of the object is fully scanned. When the second scanner 220 is scanning the side of the object to be scanned, the inclination angle of the second scanner 220 is adjusted in real time, so that the side of the object to be scanned can be fully scanned, the scanning range of the object to be scanned is expanded, and the scanning accuracy of the scanner for the image features of the object to be scanned is improved, so that the scanner can scan the image clearly, and the effect of subsequent model training is guaranteed. In the entire scanning process, the scanner is avoided from being moved over a large range, the movement trajectory of the scanner is optimized, and the scanning effect of the scanner on the image is guaranteed, which is conducive to improving the accuracy of subsequent model training. During the scanning process, the carrying unit 100 rotates in the horizontal plane, and the orientation of each side of the object to be scanned can be adjusted, so that the image features of the object can be acquired in all directions, which is conducive to improving the training efficiency of subsequent model training.
[0020] As a preferred embodiment, please refer to Figure 1 and Figure 2The carrying unit 100 includes a carrier 110, a rotation drive assembly 120 and a translation drive assembly 130. The carrier 110 is used to place an object to be scanned. The rotation drive assembly 120 is connected to the carrier 110 and is used to drive the carrier 110 to rotate in a horizontal plane. The translation drive assembly 130 is connected to the rotation drive assembly 120 and is used to drive the rotation drive assembly 120 to reciprocate in a horizontal plane. The object to be scanned is placed on the carrier 110, and the translation drive assembly 130 is started to drive the rotation drive assembly 120 to move in a straight line in a horizontal plane. The object to be scanned on the stage 110 is driven to move to the scanning unit 200, so as to shorten the distance between the object to be scanned and the first scanner 210 and the two second scanners 220, thereby improving the scanning accuracy of the first scanner 210 and the two second scanners 220 on the image features of the object to be scanned. The rotation drive component 120 is started to drive the stage 110 to rotate in a horizontal plane, so as to adjust the orientation of each side of the object to be scanned, and cooperate with the scanning work of the first scanner 210 and the two second scanners 220 to obtain the image features of the object in all directions.
[0021] As a preferred embodiment, please refer to Figure 3 The rotation drive assembly 120 includes a connecting seat 121, a sliding seat 122, a connecting frame 123 and a first rotation drive member 124. The bottom of the sliding seat 122 is fixedly connected to the top of the connecting seat 121. The connecting frame 123 is arranged above the sliding seat 122 and is fixedly connected to the bottom of the carrier 110. The first rotation drive member 124 is arranged between the sliding seat 122 and the connecting frame 123. The first rotation drive member 124 is fixedly connected to the top of the sliding seat 122. The output shaft of the first rotation drive member 124 is fixedly connected to the connecting frame 123 for driving the connecting frame 123 to rotate in a horizontal plane. When the first rotation drive member 124 is started, the connecting frame 123 is driven to rotate in a horizontal plane, and the carrier 110 is driven to rotate in a horizontal plane, so that the orientation of each side of the object to be scanned can be adjusted, and the image features of the object can be acquired in all directions in cooperation with the scanning work of the first scanner 210 and the two second scanners 220.
[0022] As a preferred embodiment, please refer to Figure 2The translation driving assembly 130 includes a base plate 131, a screw rod 132 and a second rotation driving member 133. The base plate 131 is provided with a receiving groove 1311, and the receiving groove 1311 is a long strip structure. The screw rod 132 is arranged in the receiving groove 1311 along the length direction of the receiving groove 1311. Both ends of the screw rod 132 are rotatably connected to the base plate 131. The output shaft of the second rotation driving member 133 is coaxially fixedly connected with one end of the screw rod 132, and is used to drive the screw rod 132 to rotate forward or reversely. The connecting seat 121 is slidably arranged in the receiving groove 1311. The connecting seat 121 is provided with a screw hole, and is sleeved on the screw rod 132 through the screw hole. The screw hole The sliding seat 122 is screwed to the screw rod 132, and is arranged outside the accommodating groove 1311 and is slidably connected to the top of the bottom plate 131. The second rotating driving member 133 is started, and the output shaft of the second rotating driving member 133 rotates forward, driving the screw rod 132 to rotate forward. Since the accommodating groove 1311 has a limiting effect on the connecting seat 121, when the screw rod 132 rotates forward, the connecting seat 121 can move along the length direction of the accommodating groove 1311, so as to shorten the distance between the object to be scanned and the first scanner 210 and the two second scanners 220, thereby improving the scanning accuracy of the first scanner 210 and the two second scanners 220 on the image features of the object to be scanned.
[0023] As a preferred embodiment, please refer to Figure 1 and Figure 4 The image scanning device based on image recognition model training also includes a recognition unit 300, and the recognition unit 300 is electrically connected to the first scanner 210 and the two second scanners 220 to obtain image features scanned by the first scanner 210 and the two second scanners 220, and use the image features as training samples for model training, thereby completing the training of the image recognition model.
[0024] As a preferred embodiment, please refer to Figure 1 The scanning unit 200 also includes a mounting frame 230, which is disposed at one end of the reciprocating linear motion track of the carrying unit 100. The first scanner 210 is fixedly connected to the top of the mounting frame 230, and the mounting frame 230 can provide support for the first scanner 210.
[0025] As a preferred embodiment, please refer to Figure 1 and Figure 4The scanning unit 200 further includes a spacing adjustment component 240, which is disposed on the mounting frame 230 and connected to the two second scanners 220, and is used to drive the two second scanners 220 to retract or open back to back. When the spacing adjustment component 240 is started, the two second scanners 220 are driven to retract or open back to back. When the two second scanners 220 are opened back to back, the scanning range can be expanded, so that the first scanner 210 and the two second scanners 220 can scan large objects. When the two second scanners 220 are retracted, the scanning range can be reduced, so that the first scanner 210 and the two second scanners 220 can clearly scan images when scanning small objects, thereby improving scanning accuracy.
[0026] As a preferred embodiment, please refer to Figure 4 and Figure 6 The spacing adjustment component 240 includes two swing arms 241, a bidirectional telescopic driving member 242 and two transmission members 243. The two swing arms 241 are relatively arranged on both sides of the mounting frame 230. One end of the two swing arms 241 is hinged to the mounting frame 230. The two second scanners 220 are fixedly connected to the other ends of the two swing arms 241 in a one-to-one correspondence. The bidirectional telescopic driving member 242 is horizontally arranged between the two swing arms 241 in a direction perpendicular to the reciprocating linear movement trajectory of the carrying unit 100, and is fixedly connected to the mounting frame 230. One end of the two transmission members 243 is connected to the two telescopic shafts of the bidirectional telescopic driving member 242 in a one-to-one correspondence. The other ends of the two transmission members 243 are hinged to the middle parts of the two swing arms 241 in a one-to-one correspondence, so as to convert the synchronous telescopic movement of the two telescopic axes of the bidirectional telescopic driving member 242 into the mutual retraction or back-to-back opening of the two swing arms 241. The bidirectional telescopic driving member 242 is a bidirectional cylinder. When the bidirectional telescopic driving member 242 is started, the two telescopic axes of the bidirectional telescopic driving member 242 are synchronously retracted, and the synchronous telescopic movement of the two telescopic axes of the bidirectional telescopic driving member 242 can be converted into the mutual retraction or back-to-back opening of the two swing arms 241 via the two transmission members 243, thereby realizing the mutual retraction or back-to-back opening of the two second scanners 220.
[0027] As a preferred embodiment, please refer to Figure 4 and Figure 6The two transmission members 243 each include a sliding block 2431, a first connecting rod 2432 and a second connecting rod 2433. The two sliding blocks 2431 are fixedly connected to the two telescopic shafts of the bidirectional telescopic driving member 242 in a one-to-one correspondence. Two slide grooves 231 are relatively provided on the mounting frame 230. The two slide grooves 231 extend horizontally in a direction perpendicular to the reciprocating linear movement trajectory of the bearing unit 100. The two sliding blocks 2431 are slidably connected to the two slide grooves 231 in a one-to-one correspondence. One end of the two first connecting rods 2432 is hinged to the two sliding blocks 2431 in a one-to-one correspondence. The two second connecting rods 2433 One end is hinged to the other end of the two first connecting rods 2432 in a one-to-one relationship, and the other end of the two second connecting rods 2433 is hinged to the middle part of the two swing arms 241 in a one-to-one relationship. When the bidirectional telescopic driving member 242 is started, the two telescopic axes of the bidirectional telescopic driving member 242 are synchronously telescoped, and drive the two sliding blocks 2431 to slide towards each other or away from each other in the corresponding sliding grooves 231, and then the two swing arms 241 are retracted or opened away from each other through the transmission action of the first connecting rod 2432 and the second connecting rod 2433, thereby realizing the retraction or opening of the two second scanners 220.
[0028] As a preferred embodiment, please refer to Figure 1 and Figure 5 The scanning unit 200 also includes an inclination adjustment component 250, which is connected to the bottom of the mounting frame 230 and is used to drive the mounting frame 230 to reciprocate around the rotating axis at the bottom thereof in a vertical plane to adjust the inclination of the mounting frame 230. When the inclination adjustment component 250 is started, it drives the mounting frame 230 to reciprocate around the rotating axis at the bottom thereof in a vertical plane, and can adjust the inclination of the mounting frame 230, thereby adjusting the inclination of the first scanner 210 and the two second scanners 220, thereby expanding the scanning range of the object to be scanned, improving the scanning accuracy of the scanner for the image features of the object to be scanned, enabling the scanner to scan the image clearly, and ensuring the effect of subsequent model training.
[0029] As a preferred embodiment, please refer to Figure 4 and Figure 5The tilt adjustment assembly 250 includes two bases 251, a rotating shaft 252 and a third rotating driving member 253. The two bases 251 are relatively arranged on both sides of the reciprocating linear movement track of the carrying unit 100. The rotating shaft 252 is horizontally arranged between the two bases 251 along a direction perpendicular to the reciprocating linear movement track of the carrying unit 100. The two ends of the rotating shaft 252 are rotatably connected to the two bases 251. The rotating shaft 252 is fixedly connected to the bottom of the mounting frame 230. The third rotating driving member 253 The output end of the third rotating driving member 253 is coaxially fixedly connected to one end of the rotating shaft 252, and is used to drive the rotating shaft 252 to reciprocate. When the third rotating driving member 253 is started, the output shaft of the third rotating driving member 253 rotates forward or reversely, driving the rotating shaft 252 to rotate forward or reversely, and driving the mounting frame 230 to reciprocate around the rotating shaft 252 in a vertical plane, so that the inclination angle of the mounting frame 230 can be adjusted, thereby adjusting the inclination angle of the first scanner 210 and the two second scanners 220.
[0030] The present invention also provides an image scanning method based on image recognition model training, which is applicable to the image scanning device based on image recognition model training, and comprises the following steps: Classify the objects to be scanned into large objects and small objects according to preset standards; When scanning the large object, the large object is placed on the stage 110, the spacing adjustment component 240 is started, driving the two second scanners 220 to open back to back to expand the scanning range, the translation drive component 130 is started, driving the stage 110 to move linearly in a horizontal plane to drive the large object to move to the scanning unit 200, the rotation drive component 120 is started, driving the stage 110 to rotate in a horizontal plane to adjust the orientation of each side of the large object, the first scanner 210 and the two second scanners 220 are started at the same time, the first scanner 210 scans the front of the large object, and the two second scanners 220 scan the two sides of the large object respectively. Scanning, the recognition unit 300 obtains the image features scanned by the first scanner 210 and the two second scanners 220, and uses the image features as training samples for model training to complete the training of the image recognition model. When the first scanner 210 scans the front of the large object and the second scanner 220 scans the side of the large object, the inclination adjustment component 250 is started to drive the mounting frame 230 to reciprocate around the rotation axis at its bottom in the vertical plane to adjust the inclination of the mounting frame 230, and adjust the inclination of the first scanner 210 in real time to fully scan the front of the large object, and adjust the inclination of the second scanner 220 in real time to fully scan the side of the large object; When scanning the small object, the small object is placed on the carrier 110, the spacing adjustment component 240 is started to drive the two second scanners 220 to close together to reduce the scanning range, the translation drive component 130 is started to drive the carrier 110 to move linearly in a horizontal plane to drive the small object to move to the scanning unit 200, the rotation drive component 120 is started to drive the carrier 110 to rotate in a horizontal plane to adjust the orientation of each side of the small object, the first scanner 210 and the two second scanners 220 are started at the same time, the first scanner 210 scans the front of the small object, and the two second scanners 220 scan the two sides of the small object respectively. Scanning, the recognition unit 300 obtains the image features scanned by the first scanner 210 and the two second scanners 220, and uses the image features as training samples for model training to complete the training of the image recognition model. When the first scanner 210 scans the front of the small object and the second scanner 220 scans the side of the small object, the inclination adjustment component 250 is started to drive the mounting frame 230 to reciprocate around the rotating axis at its bottom in the vertical plane, adjust the inclination of the mounting frame 230, and adjust the inclination of the first scanner 210 in real time to fully scan the front of the small object, and adjust the inclination of the second scanner 220 in real time to fully scan the side of the small object.
[0031] In order to better understand the present invention, the following Figure 1 - Figure 6 The working principle of the technical solution of the present invention is described in detail: When in use, according to the size of the object, the bidirectional telescopic driving member 242 is started, and the two telescopic shafts of the bidirectional telescopic driving member 242 are synchronously extended and retracted, and the two sliding blocks 2431 are driven to slide closer to each other or away from each other in the corresponding sliding grooves 231, and then the two swing arms 241 are retracted or opened away from each other through the transmission action of the first connecting rod 2432 and the second connecting rod 2433, thereby realizing the mutual retraction or opening of the two second scanners 220, and adjusting the distance between the two second scanners 220 to an appropriate value, and then the second rotating driving member 133 is started, and the output shaft of the second rotating driving member 133 rotates forward, driving the screw rod 132 to rotate forward. Since the receiving groove 1311 has a limiting effect on the connecting seat 121, when the screw rod 132 rotates forward, the connecting seat 121 can move along the length direction of the receiving groove 1311, and can drive the object to be scanned to move to the scanning unit 200. The first rotating driving member 124 is started to drive the connecting frame 123 to rotate in the horizontal plane, and drives the carrier 110 to rotate in the horizontal plane, so as to adjust the orientation of each side of the object to be scanned. At the same time, the first scanner 210 and the two second scanners 220 are started. The first scanner 210 scans the front of the object to be scanned, and the two second scanners 220 scan the front of the object to be scanned respectively. When the first scanner 210 is scanning the front side of the object to be scanned, when the second scanner 220 is scanning the side side of the object to be scanned, the third rotating driving member 253 is started, and the output shaft of the third rotating driving member 253 rotates forward or reversely, driving the rotating shaft 252 to rotate forward or reversely, and driving the mounting frame 230 to reciprocate around the rotating shaft 252 in the vertical plane, and the inclination angle of the mounting frame 230 can be adjusted, so that the inclination angles of the first scanner 210 and the two second scanners 220 can be adjusted. The inclination angle of the first scanner 210 is adjusted in real time, and the front side of the object to be scanned can be fully scanned. Scanning, real-time adjustment of the inclination angle of the second scanner 220 can comprehensively scan the side of the object to be scanned, expand the scanning range of the object to be scanned, improve the scanning accuracy of the scanner for the image features of the object to be scanned, enable the scanner to scan the image clearly, ensure the effect of subsequent model training, and avoid large-scale movement of the scanner during the entire scanning process, optimize the movement trajectory of the scanner, ensure the scanning effect of the scanner on the image, which is beneficial to improving the accuracy of subsequent model training. During the scanning process, the carrier 110 rotates in the horizontal plane, and the orientation of each side of the object to be scanned can be adjusted, and the image features of the object can be acquired in all directions, which is beneficial to improving the training efficiency of subsequent model training.
[0032] The image scanning device and scanning method based on image recognition model training provided by the present invention have the following beneficial effects: (1) The stage 110 can perform reciprocating linear motion in a horizontal plane, thereby shortening the distance between the object to be scanned and the first scanner 210 and the two second scanners 220, thereby improving the scanning accuracy of the image features of the object to be scanned by the first scanner 210 and the two second scanners 220. During the scanning process, the stage 110 rotates in a horizontal plane, which can adjust the orientation of each side of the object to be scanned, and can obtain the image features of the object in all directions, which is conducive to improving the training efficiency of subsequent model training; (2) During the entire scanning process, the scanner is prevented from moving over a large range, the scanner's motion trajectory is optimized, and the scanning effect of the scanner on the image is ensured, which is conducive to improving the accuracy of subsequent model training; (3) When the first scanner 210 is scanning the front side of the object to be scanned, the inclination angle of the first scanner 210 is adjusted in real time to fully scan the front side of the object to be scanned. When the second scanner 220 is scanning the side of the object to be scanned, the inclination angle of the second scanner 220 is adjusted in real time to fully scan the side of the object to be scanned, thereby expanding the scanning range of the object to be scanned, improving the scanning accuracy of the scanner for the image features of the object to be scanned, and enabling the scanner to clearly scan the image, thereby ensuring the effect of subsequent model training.
[0033] The specific implementation of the present invention described above does not constitute a limitation on the protection scope of the present invention. Any other corresponding changes and modifications made based on the technical concept of the present invention should be included in the protection scope of the claims of the present invention.
Claims
1. An image scanning device based on image recognition model training, characterized in that: include: A carrying unit, on which the object to be scanned is placed, the carrying unit can rotate in a horizontal plane to adjust the orientation of each side of the object to be scanned, and the carrying unit can also reciprocate in a linear manner in the horizontal plane to adjust the position of the object to be scanned; The scanning unit includes a first scanner and two second scanners, wherein the first scanner is arranged at one end of the reciprocating linear motion track of the carrying unit, and is used to scan the front side of the object to be scanned, and the inclination angle of the first scanner is adjustable; the two second scanners are relatively arranged on both sides of the reciprocating linear motion track of the carrying unit and are close to the first scanner, and are respectively used to scan the two side surfaces of the object to be scanned, the two second scanners can be close to each other or away from each other, and the inclination angles of the two second scanners are adjustable.
2. The image scanning device based on image recognition model training according to claim 1, characterized in that: The carrying unit includes a carrier, a rotation drive component and a translation drive component. The carrier is used to place the object to be scanned. The rotation drive component is connected to the carrier and is used to drive the carrier to rotate in a horizontal plane. The translation drive component is connected to the rotation drive component and is used to drive the rotation drive component to move reciprocatingly in a horizontal plane.
3. The image scanning device based on image recognition model training according to claim 2, characterized in that: The rotation drive assembly includes a connecting seat, a sliding seat, a connecting frame and a first rotation drive member. The bottom of the sliding seat is fixedly connected to the top of the connecting seat. The connecting frame is arranged above the sliding seat and fixedly connected to the bottom of the carrier. The first rotation drive member is arranged between the sliding seat and the connecting frame. The first rotation drive member is fixedly connected to the top of the sliding seat. The output shaft of the first rotation drive member is fixedly connected to the connecting frame for driving the connecting frame to rotate in a horizontal plane.
4. The image scanning device based on image recognition model training according to claim 3, characterized in that: The translation drive assembly includes a base plate, a screw rod and a second rotation drive member. A receiving groove is provided on the base plate, and the receiving groove is a long strip structure. The screw rod is arranged in the receiving groove along the length direction of the receiving groove. Both ends of the screw rod are rotatably connected to the base plate. The output shaft of the second rotation drive member is coaxially fixedly connected to one end of the screw rod for driving the screw rod to rotate forward or reversely. The connecting seat is slidably arranged in the receiving groove. A screw hole is provided on the connecting seat, and the connecting seat is sleeved on the screw rod through the screw hole. The screw hole is threadedly connected to the screw rod. The sliding seat is arranged outside the receiving groove and is slidably connected to the top of the base plate.
5. The image scanning device based on image recognition model training according to claim 1, characterized in that: It also includes a recognition unit, which is electrically connected to the first scanner and the two second scanners to obtain image features scanned by the first scanner and the two second scanners, and uses the image features as training samples for model training.
6. The image scanning device based on image recognition model training according to claim 1, characterized in that: The scanning unit further comprises a mounting frame, wherein the mounting frame is arranged at one end of the reciprocating linear moving track of the carrying unit, and the first scanner is fixedly connected to the top of the mounting frame.
7. The image scanning device based on image recognition model training according to claim 6, characterized in that: The scanning unit further comprises a spacing adjustment component, which is disposed on the mounting frame and connected to both of the two second scanners, and is used for driving the two second scanners to retract toward each other or to open away from each other.
8. The image scanning device based on image recognition model training according to claim 7, characterized in that: The spacing adjustment component includes two swing arms, a bidirectional telescopic driving member and two transmission members, the two swing arms are relatively arranged on both sides of the mounting frame, one end of the two swing arms are hinged to the mounting frame, the two second scanners are fixedly connected to the other ends of the two swing arms in a one-to-one correspondence, the bidirectional telescopic driving member is horizontally arranged between the two swing arms along a direction perpendicular to the reciprocating linear movement trajectory of the bearing unit, and is fixedly connected to the mounting frame, one end of the two transmission members are connected to the two telescopic shafts of the bidirectional telescopic driving member in a one-to-one correspondence, and the other end of the two transmission members are hinged to the middle parts of the two swing arms in a one-to-one correspondence, so as to convert the synchronous telescopic movement of the two telescopic shafts of the bidirectional telescopic driving member into the mutual retraction or back-to-back opening of the two swing arms.
9. The image scanning device based on image recognition model training according to claim 6, characterized in that: The scanning unit further comprises an inclination adjustment component, which is connected to the bottom of the mounting frame and is used to drive the mounting frame to reciprocate around a rotary axis at the bottom thereof in a vertical plane to adjust the inclination of the mounting frame.
10. An image scanning method based on image recognition model training, applicable to the image scanning device based on image recognition model training as claimed in any one of claims 1 to 9, characterized in that: The steps include: Classify the objects to be scanned into large objects and small objects according to preset standards; When scanning the large object, the large object is placed on the carrying unit, the two second scanners are moved away from each other to expand the scanning range, the carrying unit moves linearly in a horizontal plane to drive the large object to move to the scanning unit, the carrying unit rotates in a horizontal plane to adjust the orientation of each side of the large object, the first scanner and the two second scanners are started at the same time, the first scanner scans the front of the large object, and the two second scanners scan the two sides of the large object respectively, while the first scanner is scanning the front of the large object, the inclination angle of the first scanner is adjusted in real time to fully scan the front of the large object, and while the second scanner is scanning the side of the large object, the inclination angle of the second scanner is adjusted in real time to fully scan the side of the large object; When scanning the small object, the small object is placed on the carrying unit, the two second scanners are moved close to each other to narrow the scanning range, the carrying unit moves linearly in a horizontal plane to drive the small object to move to the scanning unit, the carrying unit rotates in a horizontal plane to adjust the orientation of each side of the small object, the first scanner and the two second scanners are started at the same time, the first scanner scans the front of the small object, and the two second scanners scan the two sides of the small object respectively, while the first scanner is scanning the front of the small object, the inclination angle of the first scanner is adjusted in real time to fully scan the front of the small object, and while the second scanner is scanning the side of the small object, the inclination angle of the second scanner is adjusted in real time to fully scan the side of the small object.
Citation Information
Patent Citations
Image scanning device for pattern recognition
CN118509533A