Dish identification system based on image detection

Abstract

The invention relates to the technical field of image detection, in particular to a dish identification system based on image detection. The dish identification system based on image detection comprises a user, a data collection and preprocessing module, a data labeling module, a training model module, a hardware acceleration module, an identification module, a system interface and user interaction module and a continuous improvement and optimization module. According to the invention, images are collected and preprocessed through the data collection and preprocessing module to obtain clear and accurate images, then dish information labeling is carried out on the images through the data labeling module, then model training is carried out through the training model module, and application is carried out through the identification module and the model. According to the method and the system, dish identification can be carried out through dish features, compared with picture comparison, a better identification effect can be achieved, and meanwhile, the continuous improvement and optimization module can carry out optimization training on the model again through feedback, so that a more accurate use effect can be achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of image detection and recognition, and in particular to a dish recognition system based on image detection. Background Art

[0002] In order to learn about dishes during sightseeing or daily life, some users often take pictures and identify dishes through image detection. However, the existing identification methods mostly compare the taken dish pictures with the saved dish pictures, and make dish judgments based on the similarity of the pictures. This method often cannot achieve a more accurate recognition effect, and it is also difficult to optimize and improve the recognition accuracy later, resulting in poor results. Summary of the invention

[0003] The purpose of the present invention is to solve the technical problems raised in the above background technology.

[0004] The present invention adopts the following technical scheme: a dish recognition system based on image detection, and the dish recognition system based on image detection includes: users, data collection and preprocessing modules, data annotation modules, training model modules, hardware acceleration modules, recognition modules, system interface and user interaction modules, and continuous improvement and optimization modules.

[0005] Preferably, the data collection and preprocessing module is used to collect image data sets with various dishes and preprocess the collected images, thereby enabling the images to be resized, noise removed, image enhanced, etc., to improve the recognition accuracy of subsequent modules.

[0006] Preferably, the data annotation module is used to annotate the images processed by the data collection and preprocessing module, and add the corresponding dish label and other relevant information to each image. Here, the dish information can correspond to the dish image for subsequent model training.

[0007] Preferably, the training model module is used for machine learning or deep learning algorithm training model, and the image and its corresponding dish label data set are input into the model, and the model parameters are optimized by back propagation so that it can accurately identify the dish image. Here, the commonly used algorithms for machine learning or deep learning algorithm training model include convolutional neural network and target detection algorithm, which can be trained by model so that it can identify dishes by dish features, and can achieve better recognition effect than picture comparison.

[0008] Preferably, the system interface and user interaction module are used to provide a user interface, allowing users to upload images of dishes, receiving user requests, and passing the images to the recognition module, displaying the recognition results, including other relevant information such as the dish name. Here, users can perform interactive operations and perform normal dish recognition and use.

[0009] Preferably, the recognition module is used to deploy and apply the model trained by the training model module through an application program or integrate the model into an existing system. Here, the recognition model trained can be used in the recognition module so that it can perform dish recognition normally.

[0010] Preferably, the continuous improvement and optimization module is used to regularly collect user feedback, data and annotations to update and expand the system's image data set, and retrain the model with new data to improve the system's recognition accuracy. At the same time, an automated process can be used to manage data updates and model training. Here, the model is repeatedly updated and trained through the continuous improvement and optimization module to improve the recognition accuracy of the model and achieve better recognition results.

[0011] Preferably, the hardware acceleration module is used to use hardware acceleration technologies such as a graphics processor or a dedicated neural network accelerator to improve the reasoning speed and real-time performance of the recognition module and the training module. Here, the hardware acceleration module is used to provide performance acceleration for the recognition module and the training module, so that the model training and dish recognition can be performed faster, achieving better use results.

[0012] Compared with the prior art, the advantages and positive effects of the present invention are:

[0013] In the present invention, images are collected and preprocessed through a data collection and preprocessing module to obtain clear and accurate images, and then the images are annotated with dish information through a data annotation module, and then the model is trained through a training model module, and applied through a recognition module and the model, so that it can identify dishes through dish features. Compared with image comparison, better recognition effect can be achieved. At the same time, the continuous improvement and optimization module can re-optimize and train the model through feedback to achieve more accurate use effect. At the same time, the hardware acceleration module can provide performance acceleration for the recognition module and the training module, so that they can perform model training and dish recognition faster and achieve better use effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The present invention proposes a flow chart of a dish recognition system based on image detection. DETAILED DESCRIPTION

[0015] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described below in conjunction with the accompanying drawings and embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0016] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways than those described herein. Therefore, the present invention is not limited to the specific embodiments of the following disclosure.

[0017] Example

[0018] See also Figure 1 The present invention provides a technical solution: a dish recognition system based on image detection, the dish recognition system based on image detection includes: a user, a data collection and preprocessing module, a data annotation module, a training model module, a hardware acceleration module, a recognition module, a system interface and user interaction module, and a continuous improvement and optimization module. The data collection and preprocessing module can collect and preprocess images to obtain clear and accurate images, and then annotate the images with dish information through the data annotation module, and then train the model through the training model module, and apply it through the recognition module and the model, so that it can recognize dishes through dish features, and can achieve better recognition effect compared with picture comparison. At the same time, the continuous improvement and optimization module can re-optimize and train the model through feedback to achieve more accurate use effect. At the same time, the hardware acceleration module can provide performance acceleration for the recognition module and the training module, so that it can perform model training and dish recognition faster and achieve better use effect.

[0019] See also Figure 1The data collection and preprocessing module is used to collect image data sets with various dishes, and preprocess the collected images so that the images can be resized, noise removed, image enhanced, etc., to improve the recognition accuracy of subsequent modules. The data annotation module is used to annotate the images processed by the data collection and preprocessing module, and add the corresponding dish labels and other related information to each image, so that the dish information can correspond to the dish image for subsequent model training. The training model module is used for machine learning or deep learning algorithm training models, and the images and their corresponding dish label data sets are input into the model. The model parameters are optimized through back propagation so that it can accurately identify the dish image. The commonly used algorithms for machine learning or deep learning algorithm training models include convolutional neural networks and target detection algorithms, which can be trained through models to enable them to identify dishes through dish features. Compared with image comparison, better recognition effects can be achieved. The system interface and user interaction module are used to provide a user interface, allowing users to upload dish images, receive user requests, and pass images to the recognition module to display the recognition results. Other results, including other relevant information such as dish names, enable users to perform interactive operations and identify and use dishes normally. The recognition module is used to deploy and apply the model trained by the training model module through an application or integrate the model into an existing system, so that the recognition model trained at the training location can be used in the recognition module, so that it can identify dishes normally. The continuous improvement and optimization module is used to regularly collect user feedback, data and annotations to update and expand the system's image data set, and use new data to retrain the model to improve the system's recognition accuracy. At the same time, an automated process can be used to manage data updates and model training. Through the continuous improvement and optimization module, the model is repeatedly updated and trained to improve the model's recognition accuracy and achieve better recognition and use effects. The hardware acceleration module is used to use hardware acceleration technologies such as graphics processors or dedicated neural network accelerators to improve the reasoning speed and real-time performance of the recognition module and training module. The hardware acceleration module provides performance acceleration for the recognition module and training module, so that they can perform model training and dish recognition faster and achieve better use effects.

[0020] Working principle: The images are collected and preprocessed through the data collection and preprocessing module to obtain clear and accurate images, and then the images are annotated with dish information through the data annotation module. The model is then trained through the training model module, and applied through the recognition module and the model, so that it can identify dishes through dish features. Compared with image comparison, it can achieve better recognition effect. At the same time, the continuous improvement and optimization module can re-optimize and train the model through feedback to achieve more accurate use effect. At the same time, the hardware acceleration module can provide performance acceleration for the recognition module and the training module, so that it can perform model training and dish recognition faster and achieve better use effect.

[0021] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A dish recognition system based on image detection, characterized in that: The dish recognition system based on image detection includes: users, data collection and preprocessing modules, data annotation modules, training model modules, hardware acceleration modules, recognition modules, system interface and user interaction modules, and continuous improvement and optimization modules.

2. The dish recognition system based on image detection according to claim 1, characterized in that: The data collection and preprocessing module is used to collect image data sets with various dishes and preprocess the collected images.

3. The dish recognition system based on image detection according to claim 1, characterized in that: The data annotation module is used to annotate the images processed by the data collection and preprocessing module, and to add corresponding dish labels and other relevant information to each image.

4. The dish recognition system based on image detection according to claim 1, characterized in that: The training model module is used for machine learning or deep learning algorithm training models, inputs images and their corresponding dish label data sets into the model, and optimizes model parameters through back propagation so that it can accurately identify dish images.

5. The dish recognition system based on image detection according to claim 1, characterized in that: The system interface and user interaction module are used to provide a user interface, allow users to upload dish images, receive user requests, and pass the images to the recognition module to display the recognition results, including other relevant information such as the dish name.

6. The dish recognition system based on image detection according to claim 1, characterized in that: The recognition module is used to deploy and apply the model trained by the training model module through an application program or integrate the model into an existing system.

7. The dish recognition system based on image detection according to claim 1, characterized in that: The continuous improvement and optimization module is used to regularly collect user feedback, data and annotations to update and expand the system's image data set, and retrain the model with new data to improve the system's recognition accuracy. At the same time, an automated process can be used to manage data updates and model training.

8. The dish recognition system based on image detection according to claim 1, characterized in that: The hardware acceleration module is used to use hardware acceleration technologies such as a graphics processor or a dedicated neural network accelerator to improve the reasoning speed and real-time performance of the recognition module and the training module.

Images