AI open platform
Through the image acquisition, classification and processing module of the AI open platform, the personalized needs of AI model development and deployment in industry businesses are solved, and the rapid implementation of AI technology in actual scenarios and model iterative optimization is achieved.
Patent Information
- Application Number
- CN202411120055.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-22
AI Technical Summary
Generalized artificial intelligence technology cannot meet the needs of personalized applications in different industries, and it lacks the experience and capabilities of AI-related model development, integration and deployment.
Provide an open AI platform, including image acquisition module, image classification module, image processing module and data storage module, through which the original images are classified, annotated, trained and output models, and the algorithm model is distributed to edge AI devices , supports verification and editing of models.
Accelerate the implementation of AI technology in actual scenarios, quickly verify application effects and continuously optimize model iterations.
Smart Images

Figure CN119006914A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image processing, and in particular to an AI open platform. Background Art
[0002] Generalized artificial intelligence technologies (such as face recognition) cannot meet the personalized application needs of different industries. Lack of experience and capabilities in AI-related model development, integration, and deployment; To sum up the above problems, we propose an AI open platform. Summary of the invention
[0003] In order to solve the above problems, the present invention provides an AI open platform to solve the existing problems.
[0004] To achieve the above object, the present invention provides the following technical solutions: An AI open platform, including an AI open platform, an industry application platform, and several edge AI devices; The image acquisition module is used to capture and obtain original images, summarize a large number of acquired original images into an image material library, and upload the image materials in the image material library to the AI open platform; The image classification module is used to classify and summarize images and to mark them online according to the types of objects in the images; The image processing module is used to use the online annotated images as training samples, train the algorithm model through the AI open platform, and then output the algorithm model through the AI open platform; The algorithm model judgment module is used to evaluate the algorithm model through the material and output the judgment result; The data storage module is used to build an image library, image training data set, and store them by category.
[0005] A method for using an AI open platform comprises the following steps: S1. Upload the image material to the AI open platform and create training samples; S2. Create an image material training dataset and use the training model for verification; S3. After verification, the industry application platform is used to import the algorithm model file downloaded from the AI open platform into the industry application platform and edit relevant attributes for the algorithm model; S4. Select the model type and send the specific algorithm model to the edge AI device. Preferably, the creation of the training samples includes: Create a dataset; Upload the data set to the pictures saved locally on the computer, and use the image classification model to classify the pictures in the current data set according to shape, behavior or state; Each image is labeled using a visual annotation tool to form the final training sample.
[0006] Preferably, the training model verification includes: Create a training model, specify its deployment mode as EZVIZ Cloud or Edge Device, and select a pre-created training dataset; Start training. The training time is related to the number of training samples. You can close the browser during the waiting process and check again after a while. After the model training is completed, you can verify the model and view the verification results; Download the model to your local computer. Preferably, the industry application platform is used to import the algorithm model file downloaded from the AI open platform into the industry application platform, edit relevant properties for the algorithm model, and after successful import, all models that have been imported into the industry application platform can be viewed in the model list. Preferably, the model is sent down, including: Select the model type to be issued and the specific algorithm model; Select the edge AI device to be delivered through the device organization tree and confirm; After confirming the submission, you can view all model delivery records in the industry application platform, including whether the delivery is successful and other status. Compared with the prior art, the present invention has the following beneficial effects: The present invention accelerates the implementation of AI technology in actual scenarios, quickly verifies application effects, and continuously performs iterative optimization of models. BRIEF DESCRIPTION OF THE DRAWINGS
[0007] The present invention is further described in detail below in conjunction with the accompanying drawings.
[0008] Figure 1 It is a schematic diagram of the structure of the present invention; Figure 2 It is a schematic diagram of the data processing flow of the present invention; Figure 3 It is a schematic diagram of the training process of the present invention; Figure 4 A schematic diagram of the process flow of issuing the model of the present invention; Figure 5 This is a schematic diagram of the monkey picture recognition process of the present invention; Figure 6 A schematic diagram of a flow chart for creating training samples of the present invention; Figure 7 The specific object of the present invention Figure 1; Figure 8 The specific object of the present invention Figure 2 ; Fig. 9 The specific object of the present invention Figure 3 ; Fig.10 The specific object of the present invention Figure 4 ; Fig.11 The specific object of the present invention Figure 5 . DETAILED DESCRIPTION
[0009] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.
[0010] Implementation Case 1: See also Figure 1-11 , an AI open platform, including an AI open platform, an industry application platform, and several edge AI devices; The image acquisition module is used to capture and obtain original images, summarize a large number of acquired original images into an image material library, and upload the image materials in the image material library to the AI open platform; The image classification module is used to classify and summarize images and to mark them online according to the types of objects in the images; The image processing module is used to use the online annotated images as training samples, train the algorithm model through the AI open platform, and then output the algorithm model through the AI open platform; The algorithm model judgment module is used to evaluate the algorithm model through the material and output the judgment result; The data storage module is used to build an image library, image training data set, and store them by category.
[0011] A method for using an AI open platform comprises the following steps: S1. Upload the image material to the AI open platform and create training samples; S11. Create training samples: Create a dataset; Upload the data set to the pictures saved locally on the computer, and use the image classification model to classify the pictures in the current data set according to shape, behavior or state; Use visual annotation tools to annotate each image and form the final training sample S2. Create an image material training dataset and use the training model for verification; S21. Training model verification: Create a training model, specify its deployment mode as EZVIZ Cloud or Edge Device, and select a pre-created training dataset; Start training. The training time is related to the number of training samples. You can close the browser during the waiting process and check again after a while. After the model training is completed, you can verify the model and view the verification results; Download the model to your local computer S3. After verification, the industry application platform is used to import the algorithm model file downloaded from the AI open platform into the industry application platform and edit relevant attributes for the algorithm model; S4. Select the model type and send the specific algorithm model to the edge AI device; S41, model issuance; Select the model type to be issued and the specific algorithm model; Select the edge AI device to be delivered through the device organization tree and confirm; After confirming the submission, you can view all model delivery records in the industry application platform, including whether the delivery is successful and other status. Preferably, the industry application platform is used to import the algorithm model file downloaded from the AI open platform into the industry application platform, edit relevant properties for the algorithm model, and after successful import, all models that have been imported into the industry application platform can be viewed in the model list.
[0012] Specific implementation case 1; Take detecting monkeys and their number as an example: The original picture materials of the monkeys are uploaded to the AI open platform, and the pictures are classified and standardized online through visual annotation tools. Then, the AI open platform performs automated model training and outputs the algorithm model. Then, the algorithm model is downloaded and sent to the edge AI device, which detects the number of monkeys in the current area in real time.
[0013] Specific implementation case 2: Take a construction site as an example: Take pictures of the construction site in advance, then upload them to the AI open platform, classify and standardize the pictures online through visual annotation tools, and then use the AI open platform to conduct automated model training and output the algorithm model. Then, download the algorithm model and send it to the edge AI device, which will identify the labeled safety helmets, work clothes, dangerous areas, fire extinguishers, etc., detect safety helmets in the current picture, identify work clothes, identify two-person operations in dangerous places, and fire extinguishers for open-flame operations, compare the scene in the current picture with the standard specifications, and verify whether the picture is correctly identified.
[0014] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. An AI open platform, characterized in that: Including AI open platform, industry application platform, and several edge AI devices; The image acquisition module is used to capture and obtain original images, summarize a large number of acquired original images into an image material library, and upload the image materials in the image material library to the AI open platform; The image classification module is used to classify and summarize images and to mark them online according to the types of objects in the images; The image processing module is used to use the online annotated images as training samples, train the algorithm model through the AI open platform, and then output the algorithm model through the AI open platform; The algorithm model judgment module is used to evaluate the algorithm model through the material and output the judgment result; The data storage module is used to build an image library, image training data set, and store them by category.
2. The method for using an AI open platform according to claim 1, characterized in that: The following steps are involved: S1. Upload the image material to the AI open platform and create training samples; S2. Create an image material training dataset and use the training model for verification; S3. After verification, the industry application platform is used to import the algorithm model file downloaded from the AI open platform into the industry application platform and edit relevant attributes for the algorithm model; S4. Select the model type and send the specific algorithm model to the edge AI device.
3. The AI open platform according to claim 2, characterized in that: The creation of the training samples includes: Create a dataset; Upload the data set to the pictures saved locally on the computer, and use the image classification model to classify the pictures in the current data set according to shape, behavior or state; Each image is labeled using a visual annotation tool to form the final training sample.
4. The AI open platform according to claim 2, characterized in that: The training model verification includes: Create a training model, specify its deployment mode as EZVIZ Cloud or Edge Device, and select a pre-created training dataset; Start training. The training time is related to the number of training samples. You can close the browser during the waiting process and check again after a while. After the model training is completed, you can verify the model and view the verification results; Download the model to your local computer.
5. The AI open platform according to claim 2, characterized in that: The industry application platform is used to import the algorithm model file downloaded from the AI open platform to the industry application platform, edit relevant properties for the algorithm model, and after the import is successful, all models that have been imported into the industry application platform can be viewed in the model list.
6. The AI open platform according to claim 2, characterized in that: The model is issued, including: Select the model type to be issued and the specific algorithm model; Select the edge AI device to be delivered through the device organization tree and confirm; After confirming the submission, you can view all model delivery records in the industry application platform, including whether the delivery is successful and other status.
