WO2022262585A1

WO2022262585A1 - Plant picture recognition method, readable storage medium, and electronic device

Info

Publication number: WO2022262585A1
Application number: PCT/CN2022/096705
Authority: WO
Inventors: 徐青松; 李青
Original assignee: 杭州睿胜软件有限公司
Priority date: 2021-06-15
Filing date: 2022-06-01
Publication date: 2022-12-22
Also published as: CN113313193A

Abstract

The present invention provides a plant picture recognition method, a readable storage medium, and an electronic device. The plant picture recognition method comprises: obtaining a part identification result of a current plant picture; identifying the category of the part identification result; and obtaining a species recognition result of the current plant picture according to the identified category of the part recognition result, and outputting the species recognition result. That is, during recognition of a plant picture, a final output recognition result is adjusted in different situations according to the category of a plant part. For example, if it is difficult to obtain relatively accurate species information from the part photographed by a user, only information of genus can be output; and if the relatively accurate species information can be obtained from the part photographed by the user, the information of species can be output in this case, such that the recognition accuracy can be improved, and the misleading and confusion brought to the user can be avoided.

Description

Plant picture recognition method, readable storage medium and electronic device

technical field

The invention relates to the field of computer technology, in particular to a plant picture recognition method, a readable storage medium and electronic equipment.

Background technique

In the field of computer technology, there are various applications (APPs) for identifying objects to be identified, such as applications for identifying plants. These applications usually receive images from users (including static images, dynamic images, and videos, etc.), and identify objects to be identified in the images based on object recognition models established by artificial intelligence technology to obtain recognition results. For example, the recognition result obtained when the object is a living thing may be the biological classification of the object to be recognized recognized by the object recognition model, for example, the taxonomic unit may be Family, Genus, or Species. In the process of plant identification, if the user takes pictures of parts such as stems, stems, and seedlings, it will be difficult to obtain more accurate species information. If the results of the identified species are directly output, it is likely to be wrong. This can mislead or confuse users.

Contents of the invention

The object of the present invention is to provide a plant picture identification method, a readable storage medium and an electronic device, so as to solve the problem of inaccurate information obtained when identifying plants.

In order to solve the above technical problems, the present invention provides a plant picture recognition method, comprising:

Obtain the part recognition result of the current plant picture;

distinguishing categories of the part recognition results; and,

According to the distinguished category of the part recognition result, the species recognition result of the current plant picture is acquired and output.

Optionally, in the plant picture recognition method, the categories of the part recognition results are differentiated based on the accuracy of plant part recognition to species.

Optionally, in the plant picture identification method, the identified categories include a first category and a second category, and the accuracy of identifying species from plant parts of the first category is lower than that of the second category. Accuracy of identification of plant parts to species.

Optionally, in the plant picture recognition method, the method of obtaining and outputting the species recognition result of the current plant picture based on the category of the part recognition result includes:

If the category of the part recognition result is the first category, the outputted species recognition result is the genus information of the plant corresponding to the current plant picture;

If the category of the part recognition result is the second category, the outputted species recognition result is the species information of the plant corresponding to the current plant picture.

Optionally, in the plant picture recognition method, the plant picture recognition method further includes:

If the obtained part recognition results are more than two and include the part recognition results whose category belongs to the second category, then according to the part recognition results whose categories belong to the second category, obtain The species identification results of the currently described plant pictures are output.

If there are more than two part recognition results obtained, then according to the category of each part recognition result, one species recognition result is correspondingly obtained;

If the obtained species identification result includes genus information and species information of the current plant picture, then output the species information.

Optionally, in the plant picture recognition method, the part recognition result is obtained through a pre-trained plant part recognition model, and the species recognition result is obtained through a pre-trained plant classification recognition model and output.

Optionally, in the plant picture recognition method, the plant part recognition model and/or the plant classification recognition model are trained using a neural network model.

The present invention also provides a readable storage medium, where a computer program is stored in the readable storage medium, and when the computer program is executed, the plant picture recognition method as described above is realized.

The present invention also provides an electronic device. The electronic device includes a processor and a memory, and a computer program is stored in the memory. When the computer program is executed by the processor, the plant picture recognition method as described above is realized.

In summary, the plant picture recognition method, readable storage medium and electronic device provided by the present invention include: obtaining the part recognition result of the current plant picture; distinguishing the categories of the part recognition result, and, according to the distinguished The category of the part recognition result is to obtain and output the species recognition result of the current plant picture. That is, when identifying plant pictures, according to the category of plant parts, the final output recognition results are adjusted in different situations. For example, if it is difficult for the user to capture more accurate species information, only the genus information can be output. , and if the user can obtain more accurate species information in the photographed part, the species information can be output at this time, which can improve the recognition accuracy and avoid misleading or confusing the user.

Description of drawings

FIG. 1 is a flow chart of a plant picture recognition method provided by an embodiment of the present invention.

detailed description

In order to make the purpose, advantages and features of the present invention clearer, the present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the drawings are all in very simplified form and not drawn to scale, and are only used to facilitate and clearly assist the purpose of illustrating the embodiments of the present invention. In addition, the structures shown in the drawings are often a part of the actual structure. In particular, each drawing needs to display different emphases, and sometimes uses different scales. It should also be understood that, unless otherwise specified or pointed out, the terms “first”, “second”, “third” and other descriptions in the specification are only used to distinguish each component, element, step, etc. in the specification, rather than It is used to express the logical relationship or sequence relationship between various components, elements, and steps.

As shown in Figure 1, an embodiment of the present invention provides a plant picture recognition method, including the following steps:

S11, acquiring the part recognition result of the current plant picture;

S12, distinguishing the categories of the part recognition results;

S13. Acquire and output the species identification result of the current plant picture according to the distinguished category of the part identification result.

Using the plant picture recognition method provided by the embodiment of the present invention, when the plant picture taken by the user is recognized, according to the category of the plant part, the final output recognition result is adjusted in different situations. For example, if the part taken by the user is difficult to obtain Accurate species information can only be output to the genus information, and if the user can obtain more accurate species information in the photographed part, then the species information can be output, so that the recognition accuracy can be improved and the user can be avoided. Mislead or cause confusion.

The above steps S11 to S13 will be further described in detail below.

In step S11, the part recognition result can be obtained through the pre-trained plant part recognition model, and the plant part recognition model can be obtained by using a neural network model to train complete images of different plants and images of various parts of different plants .

The training step of the plant part recognition model may include: obtaining a training sample set, each sample in the training sample set is marked with its part information (including stem, stem, seedling, flower, fruit, leaf, etc.); obtaining a test sample set , each sample in the test sample set is marked with its part information, wherein the test sample set is different from the training sample set; the plant part recognition model is trained based on the training sample set; based on the The test sample set tests the plant part recognition model; if the recognition accuracy does not meet the requirements, then increase the number of image samples in the training sample set, and use the updated training sample set to retrain the plant part recognition model, Until the recognition accuracy rate of the trained plant part recognition model meets the requirements. If the recognition accuracy meets the requirements, the training ends. In one embodiment, whether the training can end can be judged based on whether the recognition accuracy rate is lower than a preset accuracy rate. In this way, the trained plant part recognition model whose output accuracy rate meets the requirements can be used for object category recognition.

A certain number of image samples marked with corresponding information are acquired for each plant category, and the number of image samples prepared for each plant category may be equal or different. Usually the number of samples in the training sample set is significantly greater than the number of samples in the test sample set, for example, the number of samples in the test sample set can account for 5% to 20% of the total image sample number, while the corresponding training sample set The number of samples of can account for 80% to 95% of the total image sample number. Those skilled in the art should understand that the number of samples in the training sample set and the test sample set can be adjusted as needed.

The neural network model may be a deep convolutional neural network (CNN) or a deep residual network (Resnet). Among them, the deep convolutional neural network is a deep feedforward neural network, which uses a convolution kernel to scan the plant image, extracts the features to be identified in the plant image, and then identifies the features to be identified in the plant. In addition, in the process of recognizing plant images, the original plant images can be directly input into the deep convolutional neural network model without preprocessing the plant images. Compared with other recognition models, the deep convolutional neural network model has higher recognition accuracy and recognition efficiency. Compared with the deep convolutional neural network model, the deep residual network model adds an identity mapping layer, which can avoid the accuracy rate saturation caused by the convolutional neural network, or even decline phenomenon. The identity mapping function of the identity mapping layer in the residual network model needs to satisfy: the sum of the identity mapping function and the input of the residual network model is equal to the output of the residual network model. After the introduction of identity mapping, the residual network model has more obvious changes in the output, so the recognition accuracy and recognition efficiency of the plant part recognition model can be greatly improved.

For plants, when the identified parts are stems, stems, seedlings, etc., it will be more difficult to obtain more accurate species information; when the identified parts are flowers, fruits, leaves, etc., it is easier to obtain accurate species information, so In step S12, the classification of the part identification result may be distinguished based on the accuracy of plant part identification to species.

In this embodiment, preferably, different parts of various plants are divided into two categories: the first category and the second category, and the accuracy of identifying species from the plant parts of the first category is lower than that of the second category. Plant parts identify the accuracy of species, the plant parts of the first category may include: stems, stems and seedlings, etc., and the plant parts of the second category may include: flowers, fruits and leaves, etc. It should be understood here that The specific division of the first category and the second category does not constitute a limitation to the present application. For example, in some other embodiments, the plant parts of the first category may also include buds, flower buds, roots, etc. The parts of the species are accurately identified, or, the plant parts of the first category also include partially missing flowers, fruits, leaves and the like.

Further, in step S13 of this embodiment, the method of obtaining and outputting the species identification result of the current plant picture based on the category of the part identification result includes:

If the category of the part recognition result is the first category, the outputted species recognition result is the genus information of the plant corresponding to the current plant picture; and, if the category of the part recognition result is the second category, then The outputted species identification result is the species information of the plant corresponding to the current plant picture.

The current plant picture may include multiple parts, for example, the picture includes both flowers and stems. Therefore, in step S11, the obtained part recognition results may be more than two. Based on this, in this embodiment, further, if in step S11, more than two part recognition results are obtained, and the part recognition results whose category belongs to the second category are included, then according to the category belonging to the For the part recognition result of the second category, the species recognition result of the current plant picture is acquired and output.

For example, if the current picture of a plant is identified as including stems, leaves, and flowers, wherein the stems belong to the first category, and the leaves and flowers belong to the second category, so the species is classified according to the leaves and flowers. Recognize and output a variety of information. In this way, while ensuring the accuracy of recognition, more specific recognition results can also be fed back to the user.

In some other embodiments, if in step S11, if more than two part recognition results are obtained, then according to the category of each part recognition result, a species recognition result is obtained respectively; if the obtained species If the recognition result includes the genus information and species information of the current plant picture, the species information is output. In this way, while ensuring the recognition accuracy, more specific recognition results can also be fed back to the user.

For example, if the current plant picture is identified to include stems and flowers, wherein the stems belong to the second category, therefore, the obtained species identification result is genus information, and the flowers belong to the second category, therefore, the acquired The species identification result of is the species information, therefore, the final output is the species information based on flower part identification.

In some other embodiments, if in step S11, if more than two part recognition results are obtained, then according to the category of each part recognition result, a species recognition result is obtained respectively; if the obtained species If the recognition result includes the genus information and species information of the current plant picture, then output the species information, if the acquired recognition result only includes the genus information or species information of the current plant picture, then output The genus information or species information with the most repetitions among all the obtained species identification results. Based on this method, the accuracy of recognition can be improved.

In step S13, the species recognition result can be obtained and output through the pre-trained plant classification recognition model. The plant classification recognition model is obtained by using the neural network model to train complete images of different plants and images of various parts of different plants.

The training process of the plant classification recognition model can be similar to the training process of the plant part recognition model, training is carried out through training samples, and testing is carried out through test samples until the recognition accuracy of the plant classification recognition model is greater than or equal to the preset When the accuracy rate is reached, the training is completed. In addition, similarly, the plant classification recognition model may also be a deep convolutional neural network (CNN) or a deep residual network (Resnet).

Different from the training of the plant part recognition model, the corresponding information labeled with each image sample for the plant classification recognition model training is a plant category, which may include plant scientific names, nicknames, category names of botanical classifications, and the like. Preferably, the image samples obtained for each plant category may include as much as possible different shooting angles, different lighting conditions, and different weathers (for example, the same plant may have different forms in sunny days and rainy days), different seasons ( For example, the same plant may have different forms in different seasons), different time (for example, the same plant may have different forms in the morning and evening every day), different growth environments (for example, the same plant may grow in different forms indoors and outdoors), different geographical Location (for example, the same plant may grow differently in different geographic locations). In these cases, the corresponding information marked for each image sample may also include information such as shooting angle, illumination, weather, season, time, growth environment or geographic location of the image sample.

This embodiment also provides a readable storage medium, where a computer program is stored on the readable storage medium, and when the computer program is executed, the plant picture recognition method as described in this embodiment is implemented.

The readable storage medium may be a tangible device capable of holding and storing instructions for use by an instruction execution device, such as but not limited to an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or the above-mentioned any suitable combination. More specific examples (a non-exhaustive list) of readable storage media include: portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disk read only memory (CD-ROM), digital versatile disk (DVD), memory stick, floppy disk, mechanically encoded device, and any suitable combination of the above. The computer programs described herein may be downloaded from a readable storage medium to various computing/processing devices, or downloaded to an external computer or external storage device over a network, such as the Internet, local area network, wide area network, and/or wireless network. Additionally, a computer program for carrying out operations of the present invention may be in the form of assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or in one or more programming languages source or object code written in any combination of . The computer program can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server . In some embodiments, electronic circuits, such as programmable logic circuits, field programmable gate arrays (FPGAs) or programmable logic arrays (PLAs), can be customized by utilizing state information from a computer program that can execute computer-programmable The program instructions are read to implement various aspects of the invention.

This embodiment also provides an electronic device, the electronic device includes a memory and a processor, and a computer program is stored in the memory, and when the computer program is executed by the processor, the plant as described in this embodiment is realized. Image recognition method.

The memory may include a random access memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located far away from the processor.

The processor can be a central processing unit (Central Processing Unit, CPU), and can also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf Programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, or the like. The processor is the control center of the electronic equipment, and uses various interfaces and lines to connect various parts of the entire electronic equipment.

The electronic device may include one or more cameras for capturing still images or recording video streams, and all components for connecting these elements to each other. While the electronic devices may include full-size personal computing devices, they may alternatively include mobile computing devices capable of wirelessly exchanging data with servers over a network, such as the Internet. The electronic device may be, for example, a smartphone, or a device such as a PDA with wireless support, a tablet PC or a netbook with information available via the Internet. In another example, the electronic device may be a wearable computing system.

The electronic device may further include a communication interface and a communication bus, wherein the processor, the communication interface, and the memory complete mutual communication through the communication bus. The communication bus may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus or the like. The communication bus can be divided into an address bus, a data bus, a control bus, and the like. The communication interface is used for communication between the electronic device and other devices.

In summary, the plant picture recognition method, readable storage medium, and electronic device provided by the embodiments of the present invention include: acquiring the part recognition result of the current plant picture; distinguishing the categories of the part recognition result, and, according to the difference The category of the part recognition result is obtained, and the species recognition result of the current plant picture is obtained and output. That is, when identifying plant pictures, according to the category of plant parts, the final output recognition results are adjusted in different situations. For example, if it is difficult for the user to capture more accurate species information, only the genus information can be output. , and if the user can obtain more accurate species information in the photographed part, the species information can be output at this time, which can improve the recognition accuracy and avoid misleading or confusing the user.

In addition, it should be understood that although the present invention has been disclosed above with preferred embodiments, the above embodiments are not intended to limit the present invention. For any person skilled in the art, without departing from the scope of the technical solution of the present invention, the technical content disclosed above can be used to make many possible changes and modifications to the technical solution of the present invention, or be modified into equivalent changes, etc. effective example. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments according to the technical essence of the present invention, which do not deviate from the content of the technical solution of the present invention, still belong to the scope of protection of the technical solution of the present invention.

Claims

A plant picture recognition method, characterized in that, comprising:

Obtain the part recognition result of the current plant picture;

distinguishing categories of the part recognition results; and,

According to the distinguished category of the part recognition result, the species recognition result of the current plant picture is acquired and output.
The plant picture recognition method according to claim 1, characterized in that, based on the accuracy of identifying species from plant parts, the categories of the part recognition results are distinguished.
The method for identifying plant pictures according to claim 2, wherein the identified categories include a first category and a second category, and the accuracy of identifying species from plant parts of the first category is less than that of the second category. The accuracy with which plant parts of a category are identified to species.
The plant picture recognition method according to claim 3, wherein the method of obtaining and outputting the species recognition result of the current plant picture based on the category of the part recognition result comprises:

If the category of the part recognition result is the first category, the outputted species recognition result is the genus information of the plant corresponding to the current plant picture;

If the category of the part recognition result is the second category, the outputted species recognition result is the species information of the plant corresponding to the current plant picture.
The plant picture recognition method according to claim 3 or 4, wherein the plant picture recognition method further comprises:

If the obtained part recognition results are more than two and include the part recognition results whose category belongs to the second category, then according to the part recognition results whose categories belong to the second category, obtain The species identification results of the currently described plant pictures are output.
The plant picture recognition method according to claim 4, wherein the plant picture recognition method further comprises:

If there are more than two part recognition results obtained, then according to the category of each part recognition result, one species recognition result is correspondingly obtained;

If the obtained species identification result includes genus information and species information of the current plant picture, then output the species information.
The plant picture recognition method according to claim 1, wherein the part recognition result is obtained through a pre-trained plant part recognition model, and the species recognition result is obtained through a pre-trained plant classification recognition model and output.
The plant picture recognition method according to claim 7, characterized in that, the plant part recognition model and/or the plant classification recognition model are obtained by training with a neural network model.
A readable storage medium, characterized in that the readable storage medium stores a computer program, and when the computer program is executed, the plant picture recognition method according to any one of claims 1-8 is realized.
An electronic device, characterized in that the electronic device includes a memory and a processor, and a computer program is stored in the memory, and when the computer program is executed by the processor, any one of claims 1 to 8 is realized. The plant picture recognition method.