A visual recognition device with locally deployed AI models

Abstract

This utility model relates to the field of visual recognition device technology, and specifically discloses a visual recognition device for locally deployed AI models. Two round rod shafts are rotatably connected inside a warehouse mobile trolley. Two spur gears are fitted onto the outer surfaces of each round rod shaft. Four lifting seats are slidably connected inside the warehouse mobile trolley, and racks are fixedly connected inside each of the four lifting seats. A drive motor is fixedly connected inside the warehouse mobile trolley. The output shaft of the drive motor drives the two round rod shafts to rotate synchronously via a synchronous pulley and synchronous belt. The rotation of the two round rod shafts drives the spur gears mounted on both sides to rotate. The four spur gears mesh with the four racks, and the meshing of the spur gears and racks drives the lifting seats to adjust their height. The four synchronously adjusting lifting seats lift the recognition device to perform height adjustment operations. This synchronicity ensures that the recognition device receives uniform force during height adjustment, avoiding tilting or shaking caused by uneven force.

Description

Technical Field

[0001] This utility model relates to the field of visual recognition device technology, and in particular to a visual recognition device for locally deploying AI models. Background Technology

[0002] Locally deployed AI model-based visual recognition devices are primarily used in warehouse environments to photograph and identify various goods. This helps warehouse managers, logistics companies, and relevant regulatory departments to accurately and in real-time grasp detailed information about inventory, such as the type, quantity, storage location, and shelf life of the goods. By photographing and recognizing goods within the warehouse, the device can quickly complete tasks such as inventory counting, goods location tracking, and inbound / outbound management, contributing to improved intelligence in warehouse management, ensuring accuracy and efficiency in goods management, and guaranteeing the smooth operation of the supply chain. Its application scenarios cover various types of warehouses, including large logistics warehouses, enterprise-owned warehouses, and e-commerce warehousing centers. Currently, locally deployed AI model-based visual recognition devices typically require the following technologies in practical applications:

[0003] 1. High-resolution image acquisition component: used to acquire clear, detailed images of goods, ensuring accurate identification of labels, markings, and appearance features on the goods;

[0004] 2. Powerful computing chips: Since running AI models locally requires high computing power, it is necessary to use GPUs, FPGAs or dedicated artificial intelligence chips to quickly process image data and complete recognition tasks.

[0005] 3. Stable and durable body structure: The warehouse environment is complex and there may be collisions of goods, dust and other situations. The equipment needs to have a robust body that can adapt to different storage environments.

[0006] 4. Flexible installation and fixing structure: The installation position and angle of the device can be flexibly adjusted according to the actual situation such as warehouse layout and shelf height to ensure that the target goods can be effectively photographed;

[0007] 5. Highly efficient data storage and management system: capable of quickly storing large amounts of cargo images and recognition results data, and enabling orderly data management and convenient retrieval.

[0008] Currently, various devices and methods are used to implement warehouse photo recognition functionality. Some use fixed visual recognition devices, installed in specific locations within the warehouse, such as top shelves or corners. These devices take photos of goods below or around the warehouse from a set angle and range, suitable for continuous monitoring of goods within a fixed area. Others employ mobile visual recognition devices, similar to mobile robots, which move along preset paths within the warehouse, taking photos of goods along the way. This allows for flexible adaptation to the recognition needs of different shelf areas. Additionally, some warehouses use handheld portable visual recognition devices, where staff hold the devices to take close-up photos of the goods to be identified, often used for temporary inventory checks or locating specific items.

[0009] However, the above method has a prominent hardware structural problem: while portable visual recognition devices typically include height adjustment functions for convenient recognition, existing devices usually lack multi-point synchronous support for height adjustment. This makes it difficult to precisely control all lifting points to reach the same height when adjusting the height. This can lead to deviations in the overall height of the device, causing the camera lens to lose its ideal vertical distance and horizontal angle with the target object. Height deviations can easily result in blurred areas of the captured image of the goods, affecting the accuracy of recognition. Utility Model Content

[0010] To address the shortcomings of existing technologies, this invention provides a visual recognition device for locally deployed AI models. It solves the problem that while portable visual recognition devices typically include height adjustment functions for convenient recognition operations, existing devices often lack multi-point synchronous support for height adjustment. This makes it difficult to precisely control all lifting points to reach the same height during adjustment, potentially leading to overall height deviations. Consequently, the camera lens may not maintain the ideal vertical distance and horizontal angle with the target object, resulting in blurred areas in the captured image of the goods and affecting recognition accuracy.

[0011] To achieve the above objectives, this utility model provides the following technical solution:

[0012] A visual recognition device for locally deployed AI models includes a warehouse mobile cart. Two round rod shafts are rotatably connected inside the warehouse mobile cart, and two spur gears are sleeved on the outer surfaces of each of the two round rod shafts. Four lifting seats are slidably connected inside the warehouse mobile cart, and racks are fixedly connected inside each of the four lifting seats. The four racks mesh with the four spur gears respectively. A drive motor is fixedly connected inside the warehouse mobile cart, and synchronous pulleys are sleeved on the outer surfaces of the two round rod shafts and the drive motor. Synchronous belts are sleeved on the outer surfaces of the two sets of synchronous pulleys.

[0013] Preferably, each of the four lifting seats is provided with a rubber suction cup at its lower end, and the four rubber suction cups are slidably connected inside the warehouse mobile trolley.

[0014] Preferably, the upper ends of the four lifting seats are fixedly connected to the identification device mounting platform, and the upper end of the identification device mounting platform is fixedly connected to the identification device body.

[0015] Preferably, the identification device body has a storage module inside.

[0016] Preferably, a set of computing modules is provided inside the identification device body near the storage module.

[0017] Preferably, the outer surface of the identification device is provided with a camera lens, and a set of universal wheels is rotatably connected inside the warehouse mobile trolley.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] 1. When adjusting the height, start the drive motor installed in the warehouse moving trolley. The output shaft of the drive motor will drive two round rod shafts to rotate synchronously through the transmission of synchronous pulleys and synchronous belts. The rotation of the two round rod shafts will drive the rotation of the spur gears installed on both sides. The four spur gears will mesh with four racks respectively. Through the meshing of the spur gears and racks, the lifting seats will be raised and lowered. The four synchronously raised and lowered lifting seats will lift the identification device to adjust its height. This synchronization ensures that the identification device is subjected to uniform force during the height adjustment process, avoiding tilting or shaking caused by uneven force, ensuring the stability of the identification device during the raising and lowering process, and helping to improve the accuracy of shooting and identification.

[0020] 2. When the device is idle and requires stable placement, the drive motor can be started to drive the four lifting seats to descend synchronously. The lower ends of the four lifting seats are connected to four rubber suction cups via external threads. The upper ends of the four rubber suction cups are mounting posts with internal threads. The mounting posts are installed on the lower ends of the lifting seats by screwing them on. The four lifting seats descend synchronously, driving the rubber suction cups to descend and adhere to the ground, ensuring the stability of the device when it is idle and avoiding accidental movement caused by external force. Attached Figure Description

[0021] The above description is only an overview of the technical solution of this utility model. In order to better understand the technical means of this utility model and to implement it in accordance with the contents of the specification, the preferred embodiments of this utility model are described in detail below with reference to the accompanying drawings.

[0022] Figure 1 This is a three-dimensional structural diagram of the present invention;

[0023] Figure 2 This is an exploded view showing the connection of the identification device body of this utility model;

[0024] Figure 3 Exploded view of the storage module connection of this utility model;

[0025] Figure 4 This is an exploded view of the lifting seat connection of this utility model.

[0026] Legend: 11. Warehouse moving trolley; 12. Round rod shaft; 13. Spur gear; 14. Lifting seat; 15. Rack; 16. Drive motor; 17. Synchronous pulley; 18. Synchronous belt; 19. Rubber suction cup; 21. Identification device mounting platform; 22. Identification device body; 23. Storage module; 24. Calculation module; 25. Camera lens; 26. Universal wheel. Detailed Implementation

[0027] This application provides a visual recognition device for locally deployed AI models, which effectively solves the problem that while mobile visual recognition devices typically have a lifting adjustment function for convenient recognition operations, existing devices usually lack multi-point synchronous support lifting function, making it difficult to accurately control each lifting point to reach the same height when adjusting the height. This can lead to deviations in the overall height of the device, preventing the camera lens from maintaining the ideal vertical distance and horizontal angle with the target object. Height deviations can easily cause blurry areas in the captured images of goods, affecting the accuracy of recognition. To address this, the drive motor installed in the warehouse trolley is activated during height adjustment. The motor's output shaft drives two round rod shafts to rotate synchronously via a synchronous pulley and belt. This rotation of the two round rod shafts drives the spur gears installed on both sides to rotate. The four spur gears mesh with four racks, which in turn drive the lifting seats to adjust their height. These four synchronously adjusting lifting seats lift the recognition device to adjust its height. This synchronicity ensures that the recognition device receives uniform force during height adjustment, preventing tilting or swaying caused by uneven force, thus guaranteeing the stability of the recognition device during lifting and lowering, and improving the accuracy of both shooting and recognition. Example

[0028] like Figure 1 , Figure 2 , Figure 3 and Figure 4As shown, the technical solution in this application embodiment effectively solves the problem that, for convenient recognition operations, movable visual recognition devices typically have a lifting adjustment function. However, existing devices usually lack multi-point synchronous support lifting function, making it difficult to accurately control each lifting point to reach the same height when adjusting the height. This may lead to a deviation in the overall height of the device, causing the shooting lens to fail to maintain an ideal vertical distance and horizontal angle with the target object. Height deviation can easily cause some areas of the captured goods image to be blurry, affecting the accuracy of recognition. The overall idea is as follows: A visual recognition device with a locally deployed AI model includes a warehouse mobile cart 11. The warehouse mobile cart 11 has two round rod shafts 12 rotatably connected inside. Two spur gears 13 are sleeved on the outer surface of each of the two round rod shafts 12. The warehouse mobile cart 11 slides inside... Four lifting seats 14 are connected, and each of the four lifting seats 14 has a rack 15 fixedly connected inside. The four racks 15 mesh with four spur gears 13 respectively. A drive motor 16 is fixedly connected inside the warehouse moving trolley 11. Synchronous pulleys 17 are sleeved on the outer surfaces of the two round rod shafts 12 and the drive motor 16. Synchronous belts 18 are sleeved on the outer surfaces of the two sets of synchronous pulleys 17. Rubber suction cups 19 are provided at the lower ends of the four lifting seats 14. The four rubber suction cups 19 are slidably connected inside the warehouse moving trolley 11 for height adjustment. When the device is activated, the drive motor 16 installed inside the warehouse mobile trolley 11 is started. The output shaft of the drive motor 16 drives the two round rod shafts 12 to rotate synchronously through the transmission of the synchronous pulley 17 and the synchronous belt 18. The rotation of the two round rod shafts 12 drives the spur gears 13 installed on both sides to rotate. The four spur gears 13 mesh with the four racks 15 respectively. Through the meshing of the spur gears 13 and the racks 15, the lifting seats 14 are adjusted in height. The four synchronously adjusted lifting seats 14 lift the identification device to adjust its height. When the device is idle and needs to be placed stably, the drive motor 16 can be started to drive the four lifting seats 14 to descend synchronously. The lower end of the four lifting seats 14 is connected to four rubber suction cups 19 through external threads. The upper end of the four rubber suction cups 19 is a mounting post with internal threads. The mounting post is installed on the lower end of the lifting seat 14 by screwing the threads. The four synchronously descending lifting seats 14 drive the rubber suction cups 19 to descend and adhere to the ground, ensuring the stability of the device when it is idle and avoiding accidental movement caused by external force.

[0029] Four lifting seats 14 are fixedly connected to the upper ends of recognition device mounting platforms 21, and recognition device bodies 22 are fixedly connected to the upper ends of recognition device mounting platforms 21. The recognition device body 22 is equipped with a storage module 23. A set of computing modules 24 is set at one end of the recognition device body 22 near the storage module 23. The set of computing modules 24 installed in the recognition device body 22 are data preprocessing components, namely an image filtering unit, an image scaling and cropping unit, a computing unit, and a result output unit. Through the data preprocessing components such as the image filtering unit and the image scaling and cropping unit, in conjunction with the computing unit responsible for running the AI ​​visual recognition model, the preprocessing of the acquired images and the recognition calculation based on the AI ​​model are completed. The recognition device body 22 has multiple network interfaces. Software applications, data storage, and related operating environments can be installed and configured on the user's own local server, computer, or other local hardware devices through the interfaces to realize the function of the body deployment. When performing the shooting and recognition operation on the inventory items, the height-adjustable recognition device mounting platforms 21 will drive the recognition device body 22 to adjust its height position to adapt to the storage height of goods at different heights.

[0030] The outer surface of the identification device body 22 is provided with a camera lens 25. A set of universal wheels 26 are rotatably connected inside the warehouse mobile trolley 11. The warehouse mobile trolley 11 can move within the warehouse working environment through the universal wheels 26 installed inside it. The warehouse mobile trolley 11 is equipped with a drive source that drives the four universal wheels 26 to rotate. The drive source is a rotary motor. During the movement of the warehouse mobile trolley 11, the identification device body 22 installed on its upper end can take pictures and identify the stored items. The camera lens 25 is installed inside the identification device body 22. The camera lens 25 can capture real-time images. The identification device body 22 installed inside the identification device body 22 is used to store the images recorded by the camera lens 25.

[0031] To address the problems existing in the prior art, this utility model provides a visual recognition device for locally deployed AI models. During height adjustment, the drive motor 16 installed inside the warehouse mobile trolley 11 is activated. The output shaft of the drive motor 16 drives two round rod shafts 12 to rotate synchronously via a synchronous pulley 17 and a synchronous belt 18. The rotation of the two round rod shafts 12 drives the rotation of spur gears 13 installed on both sides. The four spur gears 13 mesh with four racks 15 respectively. Through the meshing of the spur gears 13 and racks 15, the lifting seats 14 are raised and lowered. The four synchronously raised and lowered lifting seats 14 lift the recognition device to perform height adjustment. This synchronicity ensures that the recognition device is subjected to uniform force during height adjustment, avoiding tilting or shaking caused by uneven force, ensuring the stability of the recognition device during lifting, and improving the accuracy of shooting and recognition.

[0032] Warehouse mobile trolley 11: Serves as the carrier of the entire visual recognition device, providing the installation base for other components; its interior has space for installing components such as the round rod shaft 12 and drive motor 16; the universal wheels 26 connected to the bottom allow it to move freely within the warehouse working environment; in conjunction with the drive source (rotary motor) that drives the universal wheels 26, its position can be flexibly changed, enabling the recognition device body 22 installed at the top to photograph and recognize stored items in different locations; at the same time, it provides sliding space for the lifting seat 14, and works in conjunction with the drive motor 16, synchronous pulley 17, synchronous belt 18, round rod shaft 12, spur gear 13, rack 15, and other components to realize the height adjustment function of the recognition device;

[0033] Round rod shaft 12: Rotatably connected inside the warehouse moving trolley 11, with spur gear 13 sleeved on its outer surface, and connected to drive motor 16 through synchronous pulley 17 and synchronous belt 18; it rotates synchronously under the drive of drive motor 16, thereby driving the spur gear 13 sleeved on both sides to rotate, and is the key transmission component for transmitting the power of drive motor 16 to spur gear 13 to realize the lifting adjustment of lifting seat 14;

[0034] Spur gear 13: It is sleeved on the outer surface of the round rod shaft 12 and meshes with the rack 15; it rotates under the drive of the round rod shaft 12, and through meshing with the rack 15, it converts the rotational motion of the round rod shaft 12 into the linear lifting motion of the rack 15 and the lifting seat 14 connected thereto, thereby realizing the height adjustment of the identification device.

[0035] Lifting base 14: It is slidably connected inside the warehouse mobile trolley 11 and has a rack 15 fixedly connected inside. The lifting is achieved by the meshing of the rack 15 and the spur gear 13. The upper end is fixedly connected to the identification device mounting platform 21, which drives the identification device body 22 to adjust the height position to adapt to the storage height of goods of different heights. The lower end is equipped with a rubber suction cup 19. When the device is not in use, it can be lowered by the drive motor 16 to allow the rubber suction cup 19 to adhere to the ground, ensuring that the device is placed stably and avoiding accidental movement.

[0036] Rack 15: Fixedly connected inside the lifting seat 14 and meshing with the spur gear 13; driven by the spur gear 13, it drives the lifting seat 14 to perform linear lifting motion, and is an important transmission component for realizing the height adjustment of the lifting seat 14 and the identification device.

[0037] Drive motor 16: Fixedly connected inside the warehouse mobile trolley 11, serving as a power source. Its output shaft drives the round rod shaft 12 to rotate synchronously through the synchronous pulley 17 and synchronous belt 18, thereby driving the spur gear 13, rack 15 and lifting seat 14 to move, realizing the height adjustment of the identification device; at the same time, when the device is idle, the drive motor 16 can be controlled to drive the lifting seat 14 to descend, so that the rubber suction cup 19 can be attached to the ground, ensuring the stability of the device placement;

[0038] Synchronous pulley 17: It is respectively sleeved on the outer surface of the round rod shaft 12 and the drive motor 16, and is used in conjunction with the synchronous belt 18. Its function is to accurately and smoothly transmit the rotational motion of the output shaft of the drive motor 16 to the round rod shaft 12 through the synchronous belt 18, so as to ensure that the two round rod shafts 12 can rotate synchronously, thereby ensuring the synchronous lifting of the lifting seat 14 and maintaining the stability of the identification device during the height adjustment process.

[0039] Synchronous belt 18: It is sleeved on the outer surface of the two sets of synchronous pulleys 17 and connects the drive motor 16 and the round rod shaft 12. Through the cooperation with the synchronous pulleys 17, it realizes the power transmission from the drive motor 16 to the round rod shaft 12, ensuring that the two round rod shafts 12 rotate synchronously, so that the lifting seat 14 can be raised and lowered synchronously, ensuring the stability and consistency of the height adjustment of the identification device.

[0040] Rubber suction cup 19: It is slidably connected inside the warehouse moving trolley 11 and located at the lower end of the lifting seat 14. It is screwed to the mounting column at the lower end of the lifting seat 14 through external threads. When the lifting seat 14 descends, the rubber suction cup 19 contacts the ground and adsorbs it. It uses atmospheric pressure to generate friction to ensure that the device is stably placed on the ground when idle, avoids accidental movement due to external force, and enhances the stability of the device.

[0041] Identification device mounting platform 21: It is fixedly connected to the upper end of the four lifting seats 14, providing installation support for the identification device body 22. It works in conjunction with the lifting seats 14 to drive the identification device body 22 to adjust its height to meet the needs of shooting and identifying goods of different heights.

[0042] The main body 22 of the recognition device is equipped with a storage module 23 and a computing module 24 inside, and a camera lens 25 on its outer surface. It is the core component of the entire visual recognition device and is responsible for image acquisition, data processing, storage and result output. It captures real-time images through the camera lens 25, performs preprocessing and recognition calculations on the images using the computing module 24, and stores the captured images in the storage module 23. At the same time, it realizes local deployment through the internal network interface and works with other components to complete the task of photographing and recognizing warehouse items.

[0043] Storage module 23: Located inside the recognition device body 22, it is used to store the images recorded by the camera lens 25 and the relevant data generated during the recognition process, providing data support for subsequent data analysis, query and other operations, and working together with other components in the recognition device body 22 to ensure the integrity of the visual recognition function.

[0044] The computing module 24 is located inside the recognition device body 22 near the storage module 23. As a data preprocessing component, it includes an image filtering unit, an image scaling and cropping unit, etc., and is responsible for running the AI ​​visual recognition model. It uses high-performance computing chips such as GPUs to preprocess the images captured by the camera lens 25, such as removing noise and adjusting the size. Then, it performs recognition calculations based on the AI ​​model and transmits the recognition results to the result output unit. It is the core data processing component for realizing the visual recognition function.

[0045] The camera lens 25 is set on the outer surface of the recognition device body 22. It is used to capture real-time images and acquire image information of stored items, providing raw data for subsequent data processing and recognition. It is a key component for the visual recognition device to realize the image acquisition function.

[0046] 26 omnidirectional wheels: Rotatably connected inside the warehouse mobile trolley 11, enabling the warehouse mobile trolley 11 to move freely within the warehouse working environment. In conjunction with the drive source (rotary motor), it is convenient to move the entire visual recognition device to different positions so that the recognition device body 22 can take pictures and recognize the stored items in different positions. It is an important component for realizing the device's movement function.

[0047] Working principle:

[0048] The first step involves the warehouse mobile cart 11, which is moved within the warehouse working environment by its internal casters 26. The warehouse mobile cart 11 is equipped with a drive source that rotates the four casters 26; the drive source is a rotary motor. During movement, the warehouse mobile cart 11 can photograph and identify stored items using the recognition device 22 mounted on its upper part. The recognition device 22 contains a camera lens 25 that captures real-time images. The recognition device 22 stores the images recorded by the camera lens 25. A set of computing modules 24 installed within the recognition device 22 serves as a data preprocessing component, including an image filtering unit, an image scaling and cropping unit, and a computing unit (responsible for running the AI ​​visual recognition model; due to the high computational requirements of visual recognition tasks, GPUs are often used). It possesses powerful parallel computing capabilities, which can accelerate the operation of deep learning models. The result output unit (responsible for outputting the visual recognition results in an appropriate form, with the output method being text information, such as the name and category of the recognized object) works in conjunction with data preprocessing components such as image filtering unit, image scaling and cropping unit, and computing unit responsible for running AI visual recognition model (using high-performance computing chips such as GPU) to complete the preprocessing of the acquired images and the recognition calculation based on the AI ​​model. The recognition device body 22 has multiple network interfaces, which can be connected to external devices to install and configure software applications, data storage and related operating environments on the user's own local server, computer or other local hardware devices to realize the function of body deployment. When performing the shooting and recognition operation on inventory items, the height-adjustable recognition device mounting platform 21 will drive the recognition device body 22 to adjust its height position to adapt to the storage height of goods at different heights.

[0049] The second step involves adjusting the height by activating the drive motor 16 installed inside the warehouse moving trolley 11. The output shaft of the drive motor 16 drives two round rod shafts 12 to rotate synchronously via a synchronous pulley 17 and a synchronous belt 18. The rotation of the two round rod shafts 12 drives the spur gears 13 installed on both sides to rotate. The four spur gears 13 mesh with four racks 15 respectively. Through the meshing of the spur gears 13 and racks 15, the lifting seats 14 are raised and lowered for adjustment. The four synchronously raised and lowered lifting seats 14 then lift the identification device to perform the height adjustment operation. When the device is idle and requires stable placement, the drive motor 16 can be started to drive the four lifting seats 14 to descend synchronously. The lower ends of the four lifting seats 14 are connected to four rubber suction cups 19 by external threads. The upper ends of the four rubber suction cups 19 are mounting posts with internal threads. They are installed on the lower ends of the lifting seats 14 by screwing them on. The four synchronously descending lifting seats 14 drive the rubber suction cups 19 to descend and adhere to the ground, ensuring the stability of the device when it is idle and avoiding accidental movement caused by external force.

[0050] Finally, it should be noted that the above embodiments are merely examples for clearly illustrating the present invention and are not intended to limit the implementation. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the protection scope of this invention.

Claims

1. A visual recognition device for locally deployed AI models, comprising a warehouse mobile cart (11), wherein two round rod shafts (12) are rotatably connected inside the warehouse mobile cart (11), characterized in that, Two spur gears (13) are sleeved on the outer surfaces of the two round rod shafts (12). Four lifting seats (14) are slidably connected inside the warehouse moving trolley (11). A rack (15) is fixedly connected inside each of the four lifting seats (14). The four racks (15) mesh with the four spur gears (13) respectively. A drive motor (16) is fixedly connected inside the warehouse moving trolley (11). Among them, the outer surfaces of the two round rod shafts (12) and the drive motor (16) are all fitted with synchronous pulleys (17), and the outer surfaces of the two sets of synchronous pulleys (17) are all fitted with synchronous belts (18).

2. The visual recognition device for locally deploying AI models as described in claim 1, characterized in that, Each of the four lifting seats (14) is provided with a rubber suction cup (19) at its lower end. The four rubber suction cups (19) are slidably connected inside the warehouse mobile trolley (11).

3. The visual recognition device for locally deploying AI models as described in claim 2, characterized in that, The upper ends of the four lifting seats (14) are fixedly connected to the identification device mounting platform (21); The upper end of the identification device mounting platform (21) is fixedly connected to the identification device body (22).

4. The visual recognition device for locally deploying AI models as described in claim 3, characterized in that, The identification device body (22) is equipped with a storage module (23).

5. The visual recognition device for locally deploying AI models as described in claim 4, characterized in that, A set of computing modules (24) is provided inside the main body (22) of the identification device, near the storage module (23).

6. The visual recognition device for locally deploying AI models as described in claim 5, characterized in that, The outer surface of the identification device body (22) is provided with a camera lens (25); The warehouse mobile trolley (11) is internally connected to a set of universal wheels (26).

Images