WO2018058593A1

WO2018058593A1 - Color identification method and device for target, and computer system

Info

Publication number: WO2018058593A1
Application number: PCT/CN2016/101235
Authority: WO
Inventors: 刘晓青; 伍健荣; 白向晖
Original assignee: 富士通株式会社; 刘晓青; 伍健荣; 白向晖
Priority date: 2016-09-30
Filing date: 2016-09-30
Publication date: 2018-04-05
Also published as: CN109416747B; CN109416747A

Abstract

A color identification method and device for a target, and computer system. The method comprises: classifying, on the basis of a deep neural network, the color of a target in an image, and acquiring multiple color scores of the target (101); if the largest color score of the multiple color scores of the target is greater than a first threshold, then determining the color of the target as the color corresponding to the largest color score (102); and if the largest color score of the multiple color scores of the target is not greater than the first threshold, then performing color-based counting on the basis of a color range table, and determining the color of the target according to the color-based counting result (103). The present invention improves precision and accuracy of color identification of a target.

Description

Target color recognition method, device and computer system

Technical field

The present invention relates to the field of image processing, and in particular, to a target color recognition method, apparatus, and computer system.

Background technique

At present, it is very difficult to classify or identify the color of the target in the monitoring image, especially when the target is occluded, or the target has many components of different colors, or the target is interfered by light, etc., correctly The color of the target is even more difficult.

On the other hand, the color recognition result is closely related to the object type in the image, and the main color in the image often does not give a reasonable truth in accordance with the characteristics of the human visual system. For example, in the case where the detection target is a truck in a surveillance image, the color of the truck body is white, and the color of the front is green. In this case, the human visual system recognizes the truck as green instead of the main color of the truck. --white.

It should be noted that the above description of the technical background is only for the purpose of facilitating a clear and complete description of the technical solutions of the present invention, and is convenient for understanding by those skilled in the art. The above technical solutions are not considered to be well known to those skilled in the art simply because these aspects are set forth in the background section of the present invention.

Summary of the invention

The inventors have found that since the number of samples for target color recognition is limited, there are many obstacles in the process of target color recognition, as described in the background art, therefore, the recognition accuracy and accuracy of the current target color recognition method are not high.

The embodiment of the invention provides a target color recognition method, device and computer system, which utilizes a target color classification score output by a deep neural network (DNN) to improve the accuracy and accuracy of target color recognition.

According to a first aspect of the present invention, a target color recognition method is provided, wherein the method comprises:

Color-classifying objects in the image based on a deep neural network to obtain a plurality of color scores of the target;

Determining if the largest color score of the plurality of color scores of the target is greater than the first threshold The color is the color corresponding to the maximum color score;

If the largest color score among the plurality of color scores of the target is not greater than the first threshold, the target is color-stated based on the color range table, and the color of the target is determined according to the color statistical result.

According to a second aspect of the present invention, there is provided a target color recognition device, wherein the device comprises:

a classification unit that performs color classification on a target in an image based on a depth neural network to obtain a plurality of color scores of the target;

a recognition unit that determines that a color of the target is a color corresponding to the maximum color score when a maximum color score of the plurality of color scores of the target is greater than a first threshold; a plurality of color scores at the target When the largest color score in the middle is not greater than the first threshold, the target is color-stated based on the color range table, and the color of the target is determined according to the color statistical result.

According to a third aspect of the present invention, there is provided a computer system, wherein the computer system comprises the apparatus of the aforementioned second aspect.

The beneficial effects of the embodiments of the present invention are that the accuracy and accuracy of the target color recognition can be improved by the embodiment of the present invention.

Specific embodiments of the present invention are disclosed in detail with reference to the following description and the drawings, in which <RTIgt; It should be understood that the embodiments of the invention are not limited in scope. The embodiments of the present invention include many variations, modifications, and equivalents within the scope of the appended claims.

Features described and/or illustrated with respect to one embodiment may be used in one or more other embodiments in the same or similar manner, in combination with, or in place of, features in other embodiments. .

It should be emphasized that the term "comprising" or "comprises" or "comprising" or "comprising" or "comprising" or "comprising" or "comprises"

DRAWINGS

The elements and features described in one of the figures or one embodiment of the embodiments of the invention may be combined with the elements and features illustrated in one or more other figures or embodiments. In the accompanying drawings, like reference numerals refer to the

The accompanying drawings are included to provide a further understanding of the embodiments of the invention Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art Other drawings may also be obtained from these drawings without paying for inventive labor. In the drawing:

1 is a schematic diagram of a target color recognition method of Embodiment 1;

Figure 2 is a schematic illustration of three samples of an image;

Figure 3 is a schematic diagram of color statistics of a target to determine the color of the target;

4 is a schematic diagram of a portion of an example of a color range table;

Figure 5 is a schematic view of determining the maximum color area ratio;

6 is an overall flowchart of a target color recognition method of Embodiment 1;

Figure 7 is a schematic diagram of a target color recognition device of Embodiment 2;

8 is a schematic diagram of an identification unit of a target color recognition device of Embodiment 2;

9 is a schematic diagram of a first calculation unit of an identification unit of a target color recognition device of Embodiment 2;

Figure 10 is a schematic illustration of a computer system of the third embodiment.

detailed description

The foregoing and other features of the present invention will be apparent from the The specific embodiments of the present invention are disclosed in the specification and the drawings, which are illustrated in the embodiment of the invention The invention includes all modifications, variations and equivalents falling within the scope of the appended claims. Various embodiments of the present invention will be described below with reference to the accompanying drawings. These embodiments are merely exemplary and are not limiting of the invention.

The embodiments of the present invention will be described below with reference to the accompanying drawings.

Example 1

This embodiment provides a target color recognition method, and FIG. 1 is a schematic diagram of the method. As shown in FIG. 1, the method includes:

Step 101: Perform color classification on the target in the image based on the depth neural network to obtain multiple color scores of the target.

Step 102: If the largest color score of the plurality of color scores of the target is greater than the first threshold, determining that the color of the target is a color corresponding to the maximum color score;

Step 103: If the largest color score among the plurality of color scores of the target is not greater than the first threshold, perform color statistics on the target based on the color range table, and determine the color of the target according to the color statistical result.

In the present embodiment, as described above, the color recognition result of the target is closely related to its type. because Color information is extracted using additional shapes, texture information, etc. associated with the target type, which may use deep neural networks as feature extractors and classifiers. For the principle of the deep neural network, reference may be made to the prior art, which is not described in detail in this embodiment.

In the present embodiment, when color-classifying an object in an image based on a depth neural network, the color category of the target may be defined as white, black, yellow, red, blue, or the like according to the normal color of the target object. If the number of color samples of the target is sufficient, a more complete classification model can be trained by defining a color category with a certain target type, such as a car (white, black, yellow, red, blue, etc.). During the training process, the S and V values (HSV color space) of the sample image will be randomly stretched in a small range. At the same time, methods such as adjustment, blur, flip, and noise can be randomly used to enrich the sample. Figure 2 illustrates an example of a black car in which one image on the left is the original sample image and the two images on the right are two sample images enhanced by random data.

In this embodiment, color classification of the target in the image based on the depth neural network may obtain multiple color scores of the target, if the maximum color score of the multiple color scores is greater than the first threshold (the threshold of the classifier) And determining that the color of the target is the color corresponding to the maximum color score. However, if the maximum color score of the plurality of color scores is not greater than the first threshold, according to the current depth neural network-based classification method, the color of the target cannot be recognized, or although the target can be The color is further classified, but based on the limitations of its sample, the classification results are not satisfactory.

For example, identifying a color category based on a target type can produce a more reasonable result. However, image samples with both color and target type information are very expensive, especially for monitoring images or video. Although data enhancement can help enrich the sample, its capacity is limited, and once the classifier attempts to classify a new target color, the classifier will fail or produce a completely wrong result.

In this embodiment, for system robustness, statistical methods are used to aid in analyzing the color categories of the target. If the maximum color score of the plurality of color scores is not greater than the first threshold, color statistics are performed on the target based on the color range table, and the color of the target is determined according to the color statistical result, and the classification method based on the deep neural network is solved. The problem of the color classification result of the target cannot be accurately given.

In this embodiment, step 103 (color statisticing the target based on the color range table, and determining the color of the target according to the color statistical result) can be implemented by the method of FIG. 3, as shown in FIG. 3, the method includes:

Step 301: Select a first quantity of colors according to a sequence of a plurality of color scores of the target from high to low;

Step 302: Calculate a maximum color area of the first quantity of colors in the target area according to the color range table. ratio.

Thereby, the color of the target can be determined based on the maximum color area ratio. As shown in FIG. 3, the method may further include:

Step 303: If the maximum color area ratio is greater than the second threshold, determine that the color of the target is the color having the maximum color area ratio.

In this embodiment, as shown in FIG. 3, the method may further include:

Step 304: Acquire a target area, whereby, in step 302, a maximum color area ratio of the first number of colors in the target area acquired by step 304 may be calculated according to the color range table.

In this embodiment, as described above, the target area can be obtained by detecting the target. In this embodiment, the detection method is not limited. For example, the target may be detected by the foregoing classification method based on the depth neural network to obtain the target area. The target area can also be obtained by other detection methods.

In this embodiment, as shown in FIG. 3, after the target area is acquired, the method may further include:

Step 305: Adjust the target area, and in step 302, calculate a maximum color area ratio of the first quantity of colors in the target area adjusted by step 305 according to the color range table.

In the present embodiment, the target area is adjusted, for example, contracted, and the influence of the non-ideal factors of the edge of the target area on the recognition result can be removed. However, step 305 is optional. In the ideal case, there is no non-ideal factor at the edge of the target area, and step 305 may be omitted at this time. This embodiment does not limit the method of adjustment. By this adjustment, a new target area, that is, an adjusted target area, can be obtained.

In this embodiment, the color range table is a table for characterizing the color range, and the determination of the color range is related to the color classification and color discrimination required in the actual project, and can be determined according to the standard and the actual situation. The table is not limited. Figure 4 shows a portion of an example of a color range table, which is a color threshold range table (HSV interval) used in the experiment.

Table 1

In this embodiment, the color value may be first selected from the Pantone color card according to the target object color; then the preliminary color range is obtained from the Pantone color card values; and the color range is further refined by using the color statistics of the color samples. Make the final color range table.

Let's take yellow as an example to illustrate the process of determining the yellow range:

First, according to the Pantone color table, various yellow color values (HSV or RGB, etc.) are selected, and the selection process classifies the category and the level of detail according to the desired target color. Take the color range enclosed by the dotted line shown in Figure 4 as an example:

R: Max-255Min-206;

G: Max-237Min-157;

B: Max-134Min-0.

Then, traversing the yellow color in the table, initially determining the upper and lower thresholds of the RGB color, and similarly determining the upper and lower thresholds of other colors to be classified.

Among them, the color definition of the Pantone color table is decentralized, not a continuous interval segment. The threshold obtained by traversing in this way may be excessively included (in fact, the threshold may need to be considered in stages).

In addition, the Pantone color table is mainly used to obtain an initial threshold, which needs to be corrected to obtain the correct color range interval.

For example, the color patch is output according to the upper and lower threshold ranges of the color (in the foregoing case, the number of patches is 134*80*49), and the color patch value that is obviously wrong is filtered. Or through the color transition map of the color space in advance, preliminary determination of possible problem areas, and then visual output to reduce the workload.

Finally, the yellow block is intercepted for the yellow-category sample, and the corresponding RGB color upper and lower thresholds are counted. Compared with the foregoing results, the threshold range in the foregoing results is expanded, but the correctness of the final threshold range still needs to be verified.

Through the above method, the yellow range can be obtained, and other color ranges are obtained in the same way, thereby obtaining a color range table.

Through the color range table, the color range of each color can be obtained, and according to the color range of each color, the color area of each color can be calculated.

In this embodiment, step 302 (calculating the maximum color area ratio of the first number of colors in the target area according to the color range table) may be implemented by the method of FIG. 5, as shown in FIG. 5, the method includes:

Step 501: Calculate, according to the color range table, respective areas of the first quantity of colors in the target area;

Step 502: Calculate respective color area ratios of the first quantity of colors according to respective areas of the first quantity of colors;

Step 503: The ratio of the largest color area in the color area ratio is used as the maximum color area ratio of the first number of colors in the target area.

FIG. 6 is an overall flowchart of the method according to the present embodiment. The method of the present embodiment will be described below with reference to FIG. 6 by taking the first three colors (step 301) as an example. As shown in FIG. 6, the method includes:

Step 601: target detection;

Step 602: DNN target color classification;

Step 603: Determine whether a maximum color score (referred to as a maximum score) is greater than a first threshold;

Step 604: color segmentation;

Step 605: Determine whether the maximum color area (referred to as the maximum area) is greater than a second threshold.

In step 601, target detection is performed on the monitoring image, and the target detection may be performed by using a deep neural network-based classification method, or other methods may be used for target detection, and the target area may be obtained by target detection.

In step 602, the target is color-classified using a deep neural network-based classification method (referred to as a DNN classifier), by which a plurality of color scores of the target can be obtained. This step can be implemented by the aforementioned step 101.

In step 603, it is determined whether there is a color that satisfies the color classification result, if the maximum color score is greater than the first threshold th1, that is, if the maximum color score max_S obtained from the last SoftMax layer of the DNN classifier is greater than the first threshold th1 Then, the color attribute of the target can be directly determined, that is, the color of the target is considered to be the color having the maximum color score. The first threshold th1 is a threshold used by a deep neural network based classification method, which is an empirical value, for example 0.7. If the maximum color score cannot be greater than the first threshold th1, then step 604 is entered for further processing.

In step 604, the above target is color-stated based on the color range table to determine the color of the target based on the color statistics. This step can be implemented by the foregoing step 103. The present embodiment takes the first three colors with the highest color scores from the DNN classifier as an example, and the maximum color area of the three colors in the target area can be calculated according to the color range table. Than max_A.

In one embodiment, the best color space can be selected based on the color category of the target, such as RGB, Lab, HSV. Suppose C is one of the first three colors, the color space is HSV, then the color C The area is defined by:

C_area=SUM(I(c_h_min<H<c_h_max∩c_s_min<S<c_s_max∩c_v_min<V<c_v_max)=1)

In this formula, I is an image of the target area, and c_*_* is the color range of the color C obtained from the color range table. The color range table is made according to the target object color and the Pantone color card, as described above, and will not be described again here. It can be seen from this formula that first, the pixel points satisfying the H, S, and V threshold conditions in the image I are marked as 1, and the other regions are marked as 0, and then the image matrix is summed to obtain the number of pixels satisfying the requirements. That is the C_area value.

If the remaining two colors are A and B, then the area ratio of color C is:

C_ratio=C_area/SUM(A_area,B_area,C_area)

Similarly, the area ratio of color A and color B can be obtained: A_ratio and B_ratio.

Thus, the maximum color area ratio can be obtained:

max_A=MAX(A_ratio,B_ratio,C_ratio)

That is, by step 604, the area of the selected first three colors in the target area can be obtained, and the color area ratio of each color can be obtained, thereby obtaining the maximum color area ratio.

Step 605 compares the maximum color area ratio max_A obtained in step 604 with a second threshold value th2, and further identifies the color of the target by the second threshold value th2. The second threshold value th2 is an empirical value, for example 0.5. If the maximum color area max_A is larger than the second threshold value th2, the color having the maximum color area max_A is considered to be the color of the target. Otherwise, it is considered that the color of the target cannot be recognized by the method of the embodiment, and the processing is ended.

In this embodiment, as shown in FIG. 6, optionally, the method may further include:

Step 606: Area reshaping.

This step can be implemented by the aforementioned step 305. In step 606, the target area obtained from the target detector (DNN classifier) is re-adjusted. In one embodiment, the target area can be contracted using two parameters w and h according to the target type to reduce the background pair. The impact of the test results.

For vehicles in the surveillance image, the w and h can be obtained by:

In the above formula, B_w and B_h are the width and height of the original target area.

The value is approximately [0.1 0.2], which indicates the degree of reduction relative to the original target area. For other target types, such as people, the new area preferably only retains the up cloth area, and then adjusts the target area so that the long side is no more than 80 pixels.

With the method of the embodiment, the accuracy and accuracy of the target color recognition can be improved.

Example 2

The present embodiment provides a target color recognition device. The principle of the device is similar to that of the first embodiment. Therefore, the specific implementation may refer to the implementation of the method in the first embodiment. .

7 is a schematic diagram of a target color recognition device of the present embodiment. As shown in FIG. 7, the device 700 includes: a classification unit 701 and an identification unit 702, which classifies colors of objects in an image based on a depth neural network. Obtaining a plurality of color scores of the target; the recognition unit 702 determines that the color of the target is the color corresponding to the maximum color score when the largest color score of the plurality of color scores of the target is greater than the first threshold; When the largest color score among the plurality of color scores of the target is not greater than the first threshold, the target is color-stated based on the color range table, and the color of the target is determined according to the color statistical result.

In this embodiment, the classification unit 701 can be implemented by the foregoing step 101. The identification unit 702 can be implemented by the foregoing steps 102-103, and the content thereof is incorporated herein, and details are not described herein again.

In an embodiment of the present embodiment, as shown in FIG. 8, the identification unit 702 may include: a first selection unit 801 and a first calculation unit 802, the first selection unit 801 according to the plurality of color scores of the target In a high to low order, a first number of colors are selected; the first calculating unit 802 calculates a maximum color area ratio of the first number of colors in the target area according to the color range table.

In this embodiment, as shown in FIG. 8, the identification unit 702 may further include:

The first determining unit 803 determines that the color of the target is a color having the maximum color area ratio when the maximum color area ratio is greater than the second threshold.

In this embodiment, the first selection unit 801, the first calculation unit 802, and the first determination unit 803 respectively correspond to steps 301-303, the contents of which are incorporated herein, and are not described herein again.

In this embodiment, optionally, the identifying unit 702 may further include:

The first obtaining unit 804 obtains a target area, so that the first calculating unit 802 calculates a maximum color area ratio of the first quantity of colors in the target area acquired by the first acquiring unit according to the color range table.

In this embodiment, the first obtaining unit corresponds to step 304, and the content thereof is incorporated herein, and details are not described herein again.

In this embodiment, as shown in FIG. 8 , the identification unit 702 may further include:

The second obtaining unit 805 and the adjusting unit 806, the second acquiring unit 805 acquires a target area, and the adjusting unit 806 adjusts the target area acquired by the second acquiring unit 805, so that the first calculating unit 802 calculates according to the color range table. The adjusting unit 806 adjusts a maximum color area ratio of the first number of colors in the target area after the adjustment.

In this embodiment, the second obtaining unit 805 and the adjusting unit 806 respectively correspond to steps 304-305, and the contents thereof are incorporated herein, and details are not described herein again. In addition, the second obtaining unit 805 and the first obtaining unit 804 can be combined.

FIG. 9 is a schematic diagram of an implementation of the first computing unit 802 of the present embodiment. As shown in FIG. 9 , in the present embodiment, the first computing unit 802 corresponds to FIG. 5 , and may include: a second computing unit 901 . The third calculating unit 902 and the second determining unit 903. The second calculating unit 901 calculates an area of each of the first number of colors in the target area according to the color range table; the third calculating unit 902 calculates the color of each of the first quantity of colors according to the respective areas of the first number of colors. The area ratio; the second determining unit 903 uses the maximum color area ratio of the color area ratios as the maximum color area ratio of the first number of colors in the target area.

With the apparatus of the embodiment, the accuracy and accuracy of the target color recognition can be improved.

Example 3

This embodiment also provides a computer system configured with a target color recognition device 700 as previously described.

FIG. 10 is a schematic block diagram showing the system configuration of a computer system 1000 according to an embodiment of the present invention. As shown in FIG. 10, the computer system 1000 can include a central processor 1001 and a memory 1002; the memory 1002 is coupled to the central processor 1001. It should be noted that the figure is exemplary; other types of structures may be used in addition to or in place of the structure to implement telecommunications functions or other functions.

In one embodiment, the functionality of target color recognition device 700 can be integrated into central processor 1001. Wherein, the central processing unit 1001 can be configured to implement the target color recognition method described in Embodiment 1.

For example, the central processing unit 1001 can be configured to perform control of color categorizing objects in an image based on a depth neural network to obtain a plurality of color scores of the target; if the target has a plurality of color scores The color score is greater than the first threshold, determining that the color of the target is the color corresponding to the maximum color score; if the largest color score of the plurality of color scores of the target is not greater than the first threshold, based on the color range table Color statistics are performed on the target, and the color of the target is determined according to the color statistical result.

In another embodiment, the target color recognition device 700 can be configured separately from the central processing unit 1001. For example, the target color recognition device 700 can be configured as a chip connected to the central processing unit 1001, and the target is realized by the control of the central processing unit 1001. The function of the color recognition device 700.

As shown in FIG. 10, the computer system 1000 may further include an input unit 1003, an audio processing unit 1004, a display 1005, and a power source 1006. It should be noted that the computer system 1000 does not necessarily have to include all of the components shown in FIG. 10; in addition, the computer system 1000 may also include components not shown in FIG. 10, and reference may be made to the prior art.

As shown in FIG. 10, central processor 1001, also sometimes referred to as a controller or operational control, may include a microprocessor or other processor device and/or logic device that receives input and controls each of computer system 1000. The operation of the part.

The memory 1002 may be, for example, one or more of a buffer, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable device. The above templates, thresholds, and the like can be stored, and programs for executing related information can be stored. And the central processing unit 1001 can execute the program stored by the memory 1002 to implement information storage or processing and the like. The functions of other components are similar to those of the existing ones and will not be described here. The various components of computer system 1000 may be implemented by special purpose hardware, firmware, software or a combination thereof without departing from the scope of the invention.

With the computer system of the embodiment, the accuracy and accuracy of the target color recognition can be improved.

Embodiments of the present invention also provide a computer readable program, wherein the program causes the target color recognition device or computer system to perform the method described in Embodiment 1 when the program is executed in a target color recognition device or a computer system Target color recognition method.

An embodiment of the present invention further provides a storage medium storing a computer readable program, wherein the computer readable program causes a target color recognition device or a computer system to perform the target color recognition method described in Embodiment 1.

The above apparatus and method of the present invention may be implemented by hardware or by hardware in combination with software. The present invention relates to a computer readable program that, when executed by a logic component, enables the logic component to implement the apparatus or components described above, or to cause the logic component to implement the various methods described above Or steps. The present invention also relates to a storage medium for storing the above program, such as a hard disk, a magnetic disk, an optical disk, a DVD, a flash memory, or the like.

The target color recognition method in the target color recognition device described in connection with the embodiment of the present invention may be directly It is now hardware, a software module executed by a processor, or a combination of both. For example, one or more of the functional blocks shown in Figures 7-9 and/or one or more combinations of functional blocks may correspond to various software modules of a computer program flow, or to individual hardware modules. These software modules may correspond to the respective steps shown in Figures 1, 3-6, respectively. These hardware modules can be implemented, for example, by curing these software modules using a Field Programmable Gate Array (FPGA).

The software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. A storage medium can be coupled to the processor to enable the processor to read information from, and write information to, the storage medium; or the storage medium can be an integral part of the processor. The processor and the storage medium can be located in an ASIC. The software module can be stored in the memory of the mobile terminal or in a memory card that can be inserted into the mobile terminal. For example, if a device (such as a mobile terminal) uses a larger capacity MEGA-SIM card or a large-capacity flash memory device, the software module can be stored in the MEGA-SIM card or a large-capacity flash memory device.

One or more of the functional blocks described with respect to Figures 7-9 and/or one or more combinations of functional blocks may be implemented as a general purpose processor, digital signal processor (DSP) for performing the functions described herein. An application specific integrated circuit (ASIC), field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, or any suitable combination thereof. One or more of the functional blocks described with respect to Figures 7-9 and/or one or more combinations of functional blocks may also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, multiple microprocessors One or more microprocessors in conjunction with DSP communication or any other such configuration.

The present invention has been described in connection with the specific embodiments thereof, and it should be understood by those skilled in the art that A person skilled in the art can make various modifications and changes to the invention in accordance with the principles of the invention, which are also within the scope of the invention.

Claims

A target color recognition method, wherein the method comprises:

Color-classifying objects in the image based on a deep neural network to obtain a plurality of color scores of the target;

If the largest color score of the plurality of color scores of the target is greater than the first threshold, determining that the color of the target is the color corresponding to the maximum color score;

If the largest color score among the plurality of color scores of the target is not greater than the first threshold, the target is color-stated based on the color range table, and the color of the target is determined according to the color statistical result.
The method of claim 1, wherein color statistics are performed on the target based on a color range table, including:

Selecting a first quantity of colors according to a sequence of a plurality of color scores of the target from high to low;

A maximum color area ratio of the first number of colors in the target area is calculated according to the color range table.
The method of claim 2, wherein determining the color of the target based on the color statistics results comprises:

If the maximum color area ratio is greater than the second threshold, it is determined that the color of the target is the color having the maximum color area ratio.
The method of claim 2, wherein the color statistics are performed on the target based on a color range table, further comprising:

The target area is obtained to calculate a maximum color area ratio of the first number of colors in the acquired target area according to the color range table.
The method of claim 2, wherein the color statistics are performed on the target based on a color range table, further comprising:

Get the target area;

The target area is adjusted to calculate a maximum color area ratio of the first number of colors in the adjusted target area based on the color range table.
The method of claim 2, wherein calculating a maximum color area ratio of the first number of colors in the target area according to the color range table comprises:

Calculating respective areas of the first number of colors in the target area according to the color range table;

Calculating respective color area ratios of the first number of colors according to respective areas of the first number of colors;

The ratio of the largest color area in the color area ratio is used as the maximum color area ratio of the first number of colors in the target area.
A target color recognition device, wherein the device comprises:

a classification unit that performs color classification on a target in an image based on a depth neural network to obtain a plurality of color scores of the target;

a recognition unit that determines that a color of the target is a color corresponding to the maximum color score when a maximum color score of the plurality of color scores of the target is greater than a first threshold; a plurality of color scores at the target When the largest color score in the middle is not greater than the first threshold, the target is color-stated based on the color range table, and the color of the target is determined according to the color statistical result.
The apparatus of claim 7, wherein the identification unit comprises:

a first selection unit that selects a first quantity of colors according to a sequence of a plurality of color scores of the target from high to low;

And a first calculating unit that calculates a maximum color area ratio of the first number of colors in the target area according to the color range table.
The device according to claim 8, wherein the identification unit further comprises:

a first determining unit that determines that the color of the target is a color having the maximum color area ratio when the maximum color area ratio is greater than a second threshold.
The device according to claim 8, wherein the identification unit further comprises:

a first acquiring unit that acquires a target area, so that the first calculating unit calculates a maximum color area ratio of the first number of colors in the target area acquired by the first acquiring unit according to the color range table.
The device according to claim 8, wherein the identification unit further comprises:

a second obtaining unit that acquires a target area;

And an adjustment unit that adjusts the target area, so that the first calculating unit calculates a maximum color area ratio of the first quantity of colors in the adjusted target area of the adjustment unit according to the color range table.
The apparatus of claim 8 wherein said first computing unit comprises:

a second calculating unit, which calculates an area of each of the first number of colors in the target area according to the color range table;

a third calculating unit that calculates a color area ratio of each of the first number of colors according to respective areas of the first number of colors;

a second determining unit that uses a ratio of the largest color area of the color area ratios as a maximum color area ratio of the first number of colors in the target area.
A computer system, wherein the computer system comprises the apparatus of any of claims 7-12.