CN113837039B - Fruit growth morphology visual identification method based on convolutional neural network - Google Patents
Fruit growth morphology visual identification method based on convolutional neural network Download PDFInfo
- Publication number
- CN113837039B CN113837039B CN202111067533.8A CN202111067533A CN113837039B CN 113837039 B CN113837039 B CN 113837039B CN 202111067533 A CN202111067533 A CN 202111067533A CN 113837039 B CN113837039 B CN 113837039B
- Authority
- CN
- China
- Prior art keywords
- neural network
- model
- fruit
- convolutional neural
- modules
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 235000013399 edible fruits Nutrition 0.000 title claims abstract description 58
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 29
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000000007 visual effect Effects 0.000 title claims abstract description 12
- 239000002420 orchard Substances 0.000 claims abstract description 7
- 238000012360 testing method Methods 0.000 claims abstract description 7
- 238000005070 sampling Methods 0.000 claims description 20
- 238000012549 training Methods 0.000 claims description 19
- 230000006870 function Effects 0.000 claims description 15
- 238000001514 detection method Methods 0.000 claims description 13
- 230000004913 activation Effects 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 10
- 238000011176 pooling Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 230000007423 decrease Effects 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 6
- 238000013135 deep learning Methods 0.000 abstract description 4
- 235000012055 fruits and vegetables Nutrition 0.000 description 17
- 238000011160 research Methods 0.000 description 9
- 238000004364 calculation method Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 4
- 238000005286 illumination Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 206010063385 Intellectualisation Diseases 0.000 description 2
- 230000003213 activating effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 235000020971 citrus fruits Nutrition 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000003306 harvesting Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 150000003839 salts Chemical class 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 235000013311 vegetables Nutrition 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
- G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The invention relates to the technical field of convolutional neural networks, in particular to a fruit growth morphology visual identification method based on a convolutional neural network, which comprises the following steps: s1, image acquisition: fruit images of different forms of an orchard are collected, and the images are marked; s2, image enhancement: carrying out data enhancement on the acquired image to expand a data set; s3, building a convolutional neural network model; s4, optimizing network parameters by using an SGD optimizer; and S5, detecting the test set by using the trained optimal model, and giving out a prediction frame, a category and a confidence coefficient of each target. The invention provides a method for identifying fruit growth morphology based on a deep learning technology, which has higher identification accuracy and Faster identification speed compared with the fast-RCNN and YOLO algorithms, and has fewer model parameters.
Description
Technical Field
The invention relates to the technical field of convolutional neural networks, in particular to a fruit growth morphology visual identification method based on a convolutional neural network.
Background
The fruit industry is the third largest industry after grain and vegetables in the planting industry. In recent years, the fruit industry in China is particularly rapid in development, the planting area and yield are rapidly expanded, and the fruit industry has already developed scale advantages and is continuously growing. However, most fruit picking is still mainly finished manually at present, which is time-consuming and labor-consuming and has high labor intensity; furthermore, as the population ages and agricultural labor decreases, the cost of manual picking increases accordingly, thereby affecting the market competitiveness of the fruit. Therefore, the timely and efficient harvesting of the orchard fruits and the reduction of the picking cost are particularly important. The fruit and vegetable picking robot based on machine vision can fully utilize the information sensing capability to identify and pick fruits, so that the picking efficiency is improved, the economic benefit is improved, the income of peasants is increased, and the fruit and vegetable picking robot has become a research hotspot in the field of intelligent agricultural machinery equipment at home and abroad. However, the practical products of the picking robot are few at present, and a great deal of applications are rarely available, and the main reason is that the intelligent degree is low. In view of the above, research on related technologies of fruit picking robots is developed, and the mechanical automatic intelligent picking of orchard fruits is of great practical significance.
The fruits and vegetables have various growth forms, and the picking mechanisms of the fruits and vegetables with different growth forms are different for the picking robot, so that the primary task of the operation of the fruit and vegetable picking robot is to visually identify fruits and vegetables with different growth forms, and then the robot can select a corresponding method to finish smooth picking of the fruits and vegetables with different growth forms. However, most researches at present focus on the identification problem of fruits and vegetables in a single type of growth form, but there is no systematic research on the integrated identification of fruits and vegetables in different growth forms, but the method is a necessary link which is needed to be solved finally. At present, in the existing literature at home and abroad, there is few special research on visual identification of the growth forms of fruits and vegetables, but only the judgment of the growth forms of fruits which are blocked by single overlapping and branches and leaves in the process of fruit identification research is slightly related. The university of agriculture Zhang Yajing, et al, determines whether a multi-fruit overlap condition is present by determining a single fruit area threshold, and calculating the area of each region after the acquired image is segmented. The university of Jiangsu Cai Jianrong et al judges whether the fruits overlap or not by the minimum circumscribed rectangle side length threshold a/b >1.4 in the segmented citrus fruit image. An adult at Chinese agricultural university agrees to consider that an apple fruit is divided into several different targets due to the shade of the branches and leaves if the overlapping part of the fitting circles of two or more apple fruits is greater than 1/2 of the smallest circle thereof, and calculates the apple fruit as a fruit in the form of the shade of the branches and leaves. In the research aspect of the multi-growth-form fruit and vegetable identification method of the picking robot, the application number CN201310188346.4 patent proposes that the identification is carried out from coarse to fine based on a method combining geometric calculation and regional mapping. The above documents are mostly only the judgment of the growth form of single fruits and vegetables, but are too simple; the method for identifying fruits and vegetables with multiple growth forms studied in the early stage of the team is traditional and complicated, has limited applicability, and is not a complete and mature method for identifying fruits and vegetables with different growth forms.
Disclosure of Invention
The technical problems to be solved by the invention are as follows: the network is built based on deep learning to realize identification of the growth form of the apple fruits, so that the picking robot can automatically and visually identify the growth form of the fruits, and a foundation is laid for further selecting a corresponding picking mechanism for the picking robot.
The invention adopts the technical scheme that: a fruit growth morphology visual identification method based on a convolutional neural network comprises the following steps:
s1, image acquisition: shooting fruit images of different forms of a plurality of orchards by using single reflection, marking the images, and the fruit images of different forms have four forms, including: the branch-free stem shields the single fruit, the branch-free stem shields the overlapped fruit and the branch stem shields the overlapped fruit;
s2, image enhancement: the collected fruit images are subjected to data enhancement to expand a data set, wherein the data enhancement method comprises saturation adjustment, contrast adjustment, overturning and definition adjustment, and the images are randomly processed according to the following steps of 6:2:2 is divided into a training set, a verification set and a test set;
s3, building a convolutional neural network model: the method comprises the steps that 1 image preprocessing module, 12 convolution modules, 1 space pyramid pooling layer, 4 up-sampling modules, 8 feature fusion blocks, 12C 3 modules and 1 detection module are adopted, a network firstly carries out down-sampling through the 4 convolution modules, the quality of some key information of an image is inevitably reduced after the second and third down-sampling, 9C 3 networks are added after the second and third convolution modules in order to improve the network performance, and 3C 3 networks are added after the rest convolution modules; sending the processed sample into a space pyramid pooling layer, and inserting 3C 3 networks after the pooling layer; then, starting up-sampling operation, wherein the up-sampling operation mainly comprises 2 convolution modules for adjusting the channel number, 2 up-sampling modules for enhancing semantic information, and 3C 3 networks are inserted after each up-sampling module; repeating the operations of downsampling and upsampling after upsampling is completed, and downsampling by 2 convolution modules, wherein 3C 3 networks are inserted behind each convolution module; then 2 up-sampling modules are passed, and 3C 3 networks and 3 convolution modules are inserted behind each up-sampling module, wherein the 3 convolution modules are used for adjusting the channel number; then, passing through two downsampling modules, and inserting 3C 3 networks after each downsampling module; finally, the neural network is sent to a detection module for detection, and the expression capacity of the neural network to the model is improved through a mixed activation function in 12 convolution modules;
the model detection module evaluates the merits of the training model by calculating 3 model loss values, namely classification loss and regression loss, wherein the classification loss model is divided into: classification loss of positive samples and classification loss of foreground and background predictions of positive and negative samples; the regression Loss is calculated by CIOU_Loss, the classification Loss is calculated by BECLoss and BCEWITHLogitsLoss respectively, and finally the three Loss values are added to be used as indexes for evaluating the quality of the model, the smaller the Loss value is, the more excellent the model training is, and the optimal model is obtained until the Loss value is not changed any more;
the formula of the hybrid activation function is as follows:
f(x)＝(p 1 -p 2 )x·σ[β(p 1 -p 2 )x]+p 2 x (1)
f(x)＝xσ(x) (2)
wherein σ is a sigmoid function, p1, p2 and β use three learnable parameters to adaptively adjust, wherein formula (1) is an ACON-C activation function, and formula (2) is a SiLu activation function;
compared with the ReLU function which is most widely applied at present, the formula (1) has the characteristics of unsaturation, smoothness and non-monotonicity, and has higher calculation accuracy on a deeper neural network; in order to save training time and avoid training over-fitting, the invention uses the activating function of formula (1) in a 3X 3 convolution module in downsampling and uses the activating function of formula (2) in a 1X 1 convolution module of an adjustment channel;
s4, training a convolutional neural network model: training a convolutional neural network model through a training set, optimizing weight parameters, bias parameters and batch normalization weight parameters of the convolutional neural network by using an SGD (generalized discrete Fourier transform) optimizer, calculating the gradient of a mini-batch each time, and then updating each parameter of the model towards the opposite direction of the gradient by using a learning rate, wherein the learning rate gradually decreases as the number of iterative steps increases until the model converges;
s5, identifying the growth form of the fruits: and (3) sending the test set image into the convolutional neural network optimal model trained in the step (4) for forward propagation, and returning to a specific format of the prediction frame: and (3) performing NMS operation on the center point, the length and the width, the confidence coefficient and the classification result, setting a confidence coefficient threshold value and an IOU threshold value, and changing the center point and the length and the width of the prediction frame into: and finally, storing a prediction result.
The invention has the beneficial effects that:
the identification method of the fruit growth morphology is provided based on the deep learning technology, and compared with the fast-RCNN, YOLOv3 and YOLOv4 algorithms, the method has higher identification accuracy and Faster identification speed, and the number of model parameters is also smaller; the invention enriches the technology of the current agricultural intellectualization in the research direction of identifying the growth morphology of fruits and vegetables.
Drawings
FIG. 1 is a flow chart of a visual identification method of fruit growth morphology based on convolutional neural network;
FIG. 2 is a diagram of the network architecture of the present invention;
FIG. 3 is a block diagram of the C3 network of the present invention;
FIG. 4 is a block diagram of a feature processing network of the present invention;
FIG. 5 is a graph showing the effect of recognizing the fruit growth morphology by using the morphology recognition model according to the present invention.
Detailed Description
The invention will be further described with reference to the accompanying drawings and examples, which are simplified schematic illustrations showing only the basic structure of the invention and thus showing only those constructions that are relevant to the invention.
As shown in fig. 1, the present embodiment provides a visual recognition method of fruit growth morphology based on convolutional neural network,
s1, image acquisition: shooting fruit images of different forms of a plurality of orchards by using single reverse direction, and marking the images, wherein the different forms comprise no branch shielding single fruit, no branch shielding overlapping fruit and no branch shielding overlapping fruit, classifying and marking the acquired fruit images by using marking software, wherein samples with insufficient or unclear pixel areas are not marked so as to prevent overfitting of a neural network; in the case of approaching the image edge, the object whose image edge area is less than 15% is not marked either, because its specific feature cannot be accurately discriminated.
S2, image enhancement: for identification of fruit growth forms in an orchard, as weather and illumination conditions of one day change greatly, whether a convolutional neural network can process fruit and vegetable images acquired under different illumination conditions depends on the integrity of a training data set; meanwhile, the problem that the model lacks constraint and is over-fitted due to poor approximation value caused by too little training data is considered, so that the acquired image is subjected to data enhancement in the aspects of saturation, contrast, overturning and definition in order to enhance the richness of an experimental data set;
the human visual system can sense the invariance of the color of a target object under the change of illumination conditions and imaging conditions, but the shooting equipment is extremely sensitive to the change of environment, and the shooting image and the real image have some chromatic aberration inevitably under the influence of different illumination conditions; the contrast and the saturation are adjusted to change the brightness of the image and the contrast of the bright and dark areas so as to enhance the generalization capability of the neural network, wherein the saturation is improved by 50%, and the enhancement factor of the contrast is set to be 1.5;
in order to further expand the data set, the original image is rotated 180 degrees to improve the detection performance of the neural network; because the camera is overlong in view-finding distance or an incorrect focal length is selected, and the camera is moved, the acquired image is possibly unclear, and the detection performance of the neural network is possibly affected by the blurred image; therefore, the invention randomly selects part of original images to add spiced salt noise with variance of 0.1 so as to imitate unclear images, and the robustness of the detection model is further enhanced by using the blurred images as samples; in addition, part of the original image is sharpened, the enhancement factor is set to be 0.5, the outline of the image is compensated, the edge part of the image is enhanced, the image becomes clearer, and the image after data enhancement is processed according to the following steps: 2: the proportion of 2 is randomly divided into a training set, a verification set and a test set.
S3, constructing a convolutional neural network model, wherein the constructed network totally comprises 39 processing modules, and the specific structure of the network is shown in figure 2; before the images to be trained are sent into a convolutional neural network, in order to increase the number of small targets, the robustness of the network is better, and four images are spliced in a random scaling, random cutting and random arrangement mode; considering that the invention aims to improve the recognition rate and simultaneously maintain the real-time detection efficiency, the invention refers to the depth separable convolution principle to carry out slicing operation on the image in the image preprocessing module so as to reduce the calculation of the model parameter quantity.
After the image is briefly processed, the image is sent to a feature extraction module to fit the fruit growth morphology, and the image is subjected to downsampling operation by three convolutions; according to the invention, 3C 3 networks are inserted after each convolution operation, the specific structure of the C3 networks is shown in fig. 3, the C3 module divides the feature mapping of the base layer into two parts, and then the feature mapping is combined through a cross-stage hierarchical structure, so that the calculation amount is reduced, and meanwhile, the accuracy can be ensured; meanwhile, a residual error component is added in the C3 network, so that the gradient value of back propagation between layers can be enhanced, gradient disappearance caused by network deepening is avoided, and the characteristic of finer granularity is extracted without worrying about network degradation; after downsampling, in order to convert the convolution characteristics of the images with any scale into the same dimension, the invention introduces a spatial pyramid pooling layer for processing, which not only can enable the convolution neural network to process the images with any scale, but also can avoid the network from being over-fitted.
For a small target, the pixel information is less, the information is inevitably lost in the downsampling process, and the detection performance difference of large and small objects is high; therefore, the invention carries out up-sampling immediately after the down-sampling operation, and uses two up-sampling modules to reserve the semantic information of the feature map to the greatest extent; downsampling is then continued to pass on the strong locating features of the lower layers, the specific structure of which is shown in fig. 4; if the input and the output of the three routes are nodes at the same layer, an extra edge is added in the middle for fusion, so that more features are fused while the consumption is not increased; the invention repeatedly carries out the up-sampling and down-sampling steps to deepen the network so as to better fit the morphological characteristics of the fruits; after the morphological characteristics of the fruits are extracted by using the backbone network, the fruits are sent to a detection module for detecting the verification set and calculating three model loss values and maps, wherein the loss values comprise classification loss and regression loss, and the classification loss is divided into: classification loss of positive samples and classification loss of foreground and background predictions of positive and negative samples; the regression Loss is calculated by using CIOU_Loss, the classification Loss is calculated by using BECLoss and BCEWITHLogitsLoss respectively, and finally the three Loss values are added to be used as indexes for evaluating the quality of the model.
S4, training a convolutional neural network model: training the convolutional neural network model through a training set, and optimizing the weight parameters, bias parameters and batch normalization weight parameters of the convolutional neural network through an SGD (generalized gateway) optimizer to obtain an optimal model;
s5, loading the trained optimal model, and sending the test set image into the model for forward propagation, wherein the specific format of the return prediction frame is as follows: and (3) performing NMS operation on the center point, the length and the width, the confidence coefficient and the classification result, setting a confidence coefficient threshold value and an IOU threshold value, and changing the center point and the length and the width of the prediction frame into: and finally, storing the prediction result, wherein the effect is shown in fig. 5, and the numbers behind each category in the graph represent the confidence of the corresponding category.
The invention provides a method for identifying fruit growth morphology based on a deep learning technology, which has higher identification accuracy and Faster identification speed compared with the fast-RCNN, yolov3 and Yolov4 algorithms, and has fewer model parameters; the invention enriches the technology of the current agricultural intellectualization in the research direction of identifying the growth morphology of fruits and vegetables.
With the above-described preferred embodiments according to the present invention as an illustration, the above-described descriptions can be used by persons skilled in the relevant art to make various changes and modifications without departing from the scope of the technical idea of the present invention. The technical scope of the present invention is not limited to the description, but must be determined according to the scope of claims.
Claims (3)
1. The fruit growth morphology visual identification method based on the convolutional neural network is characterized by comprising the following steps of:
s1, image acquisition: fruit images of different forms of an orchard are collected and marked, and the fruit images of different forms comprise: the branch-free stem shields the single fruit, the branch-free stem shields the overlapped fruit and the branch stem shields the overlapped fruit;
s2, image enhancement: the collected fruit images are subjected to data enhancement to expand a data set, wherein the data enhancement method comprises saturation adjustment, contrast adjustment, overturning and definition adjustment, and the images are randomly processed according to the following steps of 6:2:2 is divided into a training set, a verification set and a test set;
s3, building a convolutional neural network model: the method comprises the steps of performing downsampling on a network through 4 convolution modules, adding 9C 3 networks after a second convolution module and a third convolution module, performing downsampling processing, sending the processed result to a spatial pyramid pooling layer, and inserting 3C 3 networks after the pooling layer, wherein the 1 image preprocessing module, the 12 convolution modules, the 1 spatial pyramid pooling layer, the 4 upsampling modules, the 8 feature fusion blocks, the 12C 3 modules and the 1 detection modules are adopted; then, up-sampling operation is carried out, wherein the up-sampling operation comprises 2 convolution modules and 2 up-sampling modules, and 3C 3 networks are inserted after each up-sampling module; repeating the operations of downsampling and upsampling after upsampling is completed, downsampling is performed through 2 convolution modules, and 3C 3 networks are inserted behind each convolution module; then 2 up-sampling modules are passed, and 3C 3 networks and 3 convolution modules are inserted behind each up-sampling module; then, passing through two downsampling modules, and inserting 3C 3 networks after each downsampling module; finally, the neural network is sent to a detection module for detection, and the expression capacity of the neural network to the model is improved through a mixed activation function in 12 convolution modules;
s4, training a convolutional neural network model: training a convolutional neural network model through a training set, optimizing weight parameters, bias parameters and batch normalization weight parameters of the convolutional neural network by using an SGD (generalized discrete Fourier transform) optimizer, calculating the gradient of a mini-batch each time, and then updating each parameter of the model towards the opposite direction of the gradient by using a learning rate, wherein the learning rate gradually decreases as the number of iterative steps increases until the model converges;
s5, identifying the growth form of the fruits: and (3) sending the test set image into the convolutional neural network model trained in the step (4) for forward propagation, and returning to a specific format of the prediction frame, wherein the specific format is as follows: and (3) performing NMS operation on the center point, the length and the width, the confidence coefficient and the classification result, setting a confidence coefficient threshold value and an IOU threshold value, and changing the center point and the length and the width of the prediction frame into: and finally, storing a prediction result.
2. The visual recognition method of fruit growth morphology based on convolutional neural network of claim 1, wherein the formula of the mixed activation function of step S3 is:
f(x)＝(p 1 -p 2 )x·σ[β(p 1 -p 2 )x]+p 2 x (1)
f(x)＝xσ(x) (2)
wherein σ is a sigmoid function, p 1 Three learnable parameters are used for self-adaptive adjustment, wherein, formula (1) is ACON-C activation function, and formula (2) is SiLu activation function; wherein the activation function of formula (1) is used in a 3 x 3 convolution module in downsampling, and the activation function of formula (2) is used in a 1 x 1 convolution module in the adjustment channel.
3. The visual recognition method of fruit growth morphology based on convolutional neural network according to claim 1, wherein the detection module evaluates the merits of the training model by calculating a classification Loss model and a regression Loss model, the classification Loss model is divided into classification Loss of positive samples and classification Loss of foreground and background predictions of the positive and negative samples, the regression Loss model is calculated by using CIOU_loss, the classification Loss model is calculated by using BECLoss and BCEWITHLogitLoss respectively, and the three Loss values are added to be used as indexes for evaluating the merits of the model until the Loss values are no longer changed to obtain an optimal model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111067533.8A CN113837039B (en) | 2021-09-13 | 2021-09-13 | Fruit growth morphology visual identification method based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111067533.8A CN113837039B (en) | 2021-09-13 | 2021-09-13 | Fruit growth morphology visual identification method based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113837039A CN113837039A (en) | 2021-12-24 |
| CN113837039B true CN113837039B (en) | 2023-10-24 |
Family
ID=78959224
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111067533.8A Active CN113837039B (en) | 2021-09-13 | 2021-09-13 | Fruit growth morphology visual identification method based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113837039B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116109956A (en) * | 2023-04-12 | 2023-05-12 | 安徽省空安信息技术有限公司 | Unmanned aerial vehicle self-adaptive zooming high-precision target detection intelligent inspection method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111784671A (en) * | 2020-06-30 | 2020-10-16 | 天津大学 | Pathological image focus region detection method based on multi-scale deep learning |
| CN113052210A (en) * | 2021-03-11 | 2021-06-29 | 北京工业大学 | Fast low-illumination target detection method based on convolutional neural network |
| WO2021139069A1 (en) * | 2020-01-09 | 2021-07-15 | 南京信息工程大学 | General target detection method for adaptive attention guidance mechanism |
| US11074711B1 (en) * | 2018-06-15 | 2021-07-27 | Bertec Corporation | System for estimating a pose of one or more persons in a scene |
-
2021
- 2021-09-13 CN CN202111067533.8A patent/CN113837039B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11074711B1 (en) * | 2018-06-15 | 2021-07-27 | Bertec Corporation | System for estimating a pose of one or more persons in a scene |
| WO2021139069A1 (en) * | 2020-01-09 | 2021-07-15 | 南京信息工程大学 | General target detection method for adaptive attention guidance mechanism |
| CN111784671A (en) * | 2020-06-30 | 2020-10-16 | 天津大学 | Pathological image focus region detection method based on multi-scale deep learning |
| CN113052210A (en) * | 2021-03-11 | 2021-06-29 | 北京工业大学 | Fast low-illumination target detection method based on convolutional neural network |
Non-Patent Citations (2)
| Title |
|---|
| 基于具有空间注意力机制的Mask R-CNN的口腔白斑分割;谢飞;穆昱;管子玉;沈雪敏;许鹏飞;王和旭;;西北大学学报(自然科学版)(第01期);全文 * |
| 基于深度学习的水果采摘机器人视觉识别系统设计;伍锡如;雪刚刚;刘英璇;;农机化研究(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113837039A (en) | 2021-12-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jia et al. | Detection and segmentation of overlapped fruits based on optimized mask R-CNN application in apple harvesting robot | |
| CN110619632B (en) | Mango example confrontation segmentation method based on Mask R-CNN | |
| CN109325960B (en) | Infrared cloud chart cyclone analysis method and analysis system | |
| CN113392775B (en) | Sugarcane seedling automatic identification and counting method based on deep neural network | |
| CN110059586B (en) | Iris positioning and segmenting system based on cavity residual error attention structure | |
| CN111179216B (en) | Crop disease identification method based on image processing and convolutional neural network | |
| CN110414509B (en) | Port docking ship detection method based on sea-land segmentation and characteristic pyramid network | |
| CN110569747A (en) | method for rapidly counting rice ears of paddy field rice by using image pyramid and fast-RCNN | |
| CN111611972B (en) | Crop leaf type identification method based on multi-view multi-task integrated learning | |
| CN111860330A (en) | Apple leaf disease identification method based on multi-feature fusion and convolutional neural network | |
| Lv et al. | A visual identification method for the apple growth forms in the orchard | |
| CN112381764A (en) | Crop disease and insect pest detection method | |
| CN111723764A (en) | Improved fast RCNN hydroponic vegetable seedling state detection method | |
| CN113887517B (en) | Crop remote sensing image semantic segmentation method based on parallel attention mechanism | |
| CN114863263B (en) | Snakehead fish detection method for blocking in class based on cross-scale hierarchical feature fusion | |
| CN113657326A (en) | Weed detection method based on multi-scale fusion module and feature enhancement | |
| CN110969121A (en) | High-resolution radar target recognition algorithm based on deep learning | |
| CN113837039B (en) | Fruit growth morphology visual identification method based on convolutional neural network | |
| Gao et al. | Recognition and Detection of Greenhouse Tomatoes in Complex Environment. | |
| Zhang et al. | Robust image segmentation method for cotton leaf under natural conditions based on immune algorithm and PCNN algorithm | |
| CN112380917A (en) | A unmanned aerial vehicle for crops plant diseases and insect pests detect | |
| Kundur et al. | Deep convolutional neural network architecture for plant seedling classification | |
| Asriny et al. | Transfer learning VGG16 for classification orange fruit images | |
| CN109102019A (en) | Image classification method based on HP-Net convolutional neural networks | |
| CN114037737B (en) | Neural network-based offshore submarine fish detection and tracking statistical method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |