Underwater video fish identification method, system and storage medium

Abstract

The present invention discloses an underwater video fish recognition method, system and storage medium, wherein the recognition method comprises: S01, constructing a fish image database; S02, acquiring underwater video images; S03, dividing the underwater video images into image frame files, then importing the files into an image recognition system for recognition processing, and outputting detection image frames with box selection marks; S04, extracting areas with box selection marks in the detection image frames, importing the areas into a fish image database for matching processing, and outputting image matching results; S05, when the matching results meet preset conditions, outputting fish information; the scheme is reliable to implement, has high processing efficiency and good user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of biological image recognition, and in particular to an underwater video fish recognition method, system and storage medium. Background Art

[0002] With the increasing demand for marine resources, the development of marine fishery resources has received more and more attention. Fishery resource survey and assessment is an important basis for the rational development of marine fishery resources, which urgently requires the ability to quickly identify fish species.

[0003] The traditional method of fish resource survey is mainly based on trawling. Compared with traditional trawling identification methods, the use of acoustic induction methods has been increasingly popularized in recent years. Acoustic methods have the advantages of being fast, convenient and having a low risk of damage to biological resources. However, the use of acoustic methods will still interfere with fish populations to a certain extent. With the increasing improvement of current underwater photography technology, if it is possible to directly use video recording combined with image recognition to assist in fish identification, then the interference to fish can be minimized. At the same time, massive data can be collected from cloud servers for identification and recording. Summary of the invention

[0004] In view of this, the purpose of the present invention is to propose an underwater video fish identification method, system and storage medium with reliable implementation, high processing efficiency and good user experience.

[0005] In order to achieve the above technical objectives, the technical solution adopted by the present invention is:

[0006] A method for identifying fish from underwater videos, comprising:

[0007] S01, build fish image database;

[0008] S02, obtaining underwater video images;

[0009] S03, dividing the underwater video image into image frame files, and then importing the files into the image recognition system for recognition processing, and outputting the detection image frames with box selection marks;

[0010] S04, extracting the area with the box selection mark in the detection image frame, importing it into the fish image database for matching processing, and outputting the image matching result;

[0011] S05. When the image matching result meets the preset conditions, the fish information is output.

[0012] Wherein, as a possible implementation manner, further, the detection image frame with a frame selection mark is to frame and mark the whole fish, fish head, fish body, fish fins and fish tail in the image frame respectively;

[0013] The data stored in the fish image database includes multi-view overall image data of different fish and local image data of the fish head, body, fins and tail that are segmented and stored separately from multiple viewpoints.

[0014] As a preferred implementation option, preferably, the image matching result is also normalized to obtain a 1×2 matrix data result (x1, y1), where x1+y1=1, x1 is the matching probability, and y1 is the matching probability.

[0015] As a preferred implementation option, preferably, the method for segmenting underwater video images into image frame files is:

[0016] The video stream in the underwater video image is converted into image frames, and then the corresponding image frames are extracted at preset time intervals to generate detection image frames, which are used to be imported into the image recognition system for recognition processing.

[0017] In terms of matching processing, different matching mechanisms have different efficiencies. Therefore, this solution also provides two data matching mechanisms, as follows:

[0018] As one of the matching mechanisms, preferably, the fish data in the fish image database is further classified according to the body shape proportion characteristics and the head characteristics of the fish and an index library is established, wherein the body shape proportion characteristics are determined according to the length of the fish and the height ratio range from its abdomen to the back, and the body shape proportion characteristic value interval includes at least: [0, 1], [1, 2], [2, 5], [5, 10], [10, 15], [15, 30], [30, +∞];

[0019] When the detection image frame is imported into the fish image database for matching, an edge detection algorithm is used to extract the overall outline of the fish in the detection image frame, the height and length of the fish are estimated, and the length-to-height ratio is calculated, and the length-to-height ratio is matched to the index library of the fish image database, and then matching is performed in the data set under the corresponding body shape proportion feature value interval, and fish data with the highest probability of matching the fish in the detection image frame is obtained in real time, and the matching probability of the fish data is set as the first data;

[0020] When the first data is greater than a first preset value, it is outputted as a temporary image matching result first, and then when it meets a preset condition, the fish information is outputted;

[0021] When the first data is smaller than the second preset value, continue to call the data in the fish database for matching until the first data is larger than the second preset value or the traversal and matching of the fish database is completed;

[0022] Among them, after the matching of the data set corresponding to the body shape proportion feature value interval is completed, the data sets under the remaining body shape proportion feature value intervals are matched in order from small to large according to the difference between the starting value of the body shape proportion feature value interval and the length-to-height ratio of the fish in the current detection image frame.

[0023] As a preferred implementation option, preferably, the edge detection algorithm is a Canny edge detection algorithm or a Scharr edge detection algorithm.

[0024] As the second matching mechanism, preferably, when the detection image frame is imported into the fish image database for matching, a detection neural network algorithm is also used to judge the area with a box selection mark in the detection image frame, and then the detection image frame is classified into a first detection set with fish head image information, a second detection set with fish body information but without fish head, and a third detection set without fish head and without fish body according to whether the fish head and fish body are reflected;

[0025] in,

[0026] The detection image in the first detection set uses a traversal method to match the area marked with fish heads in the detection image frame with the fish head data in the fish image database. When the matching probability is greater than a preset value, the area marked with the whole fish in the detection image frame is extracted for further matching with the multi-view overall image data of the fish corresponding to the fish head data, and the fish data with the highest probability of matching the fish in the detection image frame is obtained in real time, which is set as the second data;

[0027] When the second data is greater than the first preset value, it is output as a temporary image matching result first, and then the fish information is output when it meets the preset conditions.

[0028] When the second data is less than the second preset value, continue to call the data in the fish database for matching until the second data is greater than the second preset value or the traversal and matching of the fish database is completed;

[0029] The detection image in the second detection set uses a traversal method to match the area marked with the fish body in the detection image frame with the fish body data in the fish image database. When the matching probability is greater than a preset value, the area marked with the whole fish in the detection image frame is extracted for further matching with the multi-view overall image data of the fish corresponding to the fish body data, and the fish data with the highest probability of matching the fish in the detection image frame is obtained in real time, which is set as the third data;

[0030] When the third data is greater than the first preset value, it is outputted as a temporary image matching result first, and then the fish information is outputted when it meets the preset conditions.

[0031] When the third data is less than the second preset value, continue to call the data in the fish database for matching until the third data is greater than the second preset value or the traversal and matching of the fish database is completed;

[0032] The detection image in the third detection set uses a traversal method to match the area marked with the whole fish in the detection image frame with the multi-view whole image data of the fish in the fish image database, and obtains the fish data with the highest probability of matching the fish in the detection image frame in real time, which is set as the fourth data;

[0033] When the fourth data is greater than the first preset value, it is outputted as a temporary image matching result first, and then the fish information is outputted when it meets the preset conditions.

[0034] When the fourth data is smaller than the second preset value, the calling of the data in the fish database for matching continues until the fourth data is larger than the second preset value or the traversal and matching of the fish database is completed.

[0035] As a preferred implementation option, preferably, the fish image database also stores fish skin disease image data and fish abnormal posture image data, and the output image matching results include: fish information, skin disease information and abnormal state information.

[0036] As a preferred implementation option, preferably, when the matching result meets the preset conditions, the fish information is output in a visual manner;

[0037] Among them, the method of outputting fish information in a visual manner includes: drawing a marking line for the entire area of ​​the fish marked by the box selection in the detection image frame in real time, and outputting the fish information in text at the end of the marking line.

[0038] Based on the above method, this solution also provides an underwater video fish identification system, including:

[0039] A storage unit, used for storing a fish image database;

[0040] An image acquisition unit, used to acquire underwater video;

[0041] An image processing unit, used for segmenting underwater video into image frame files;

[0042] An image recognition unit is used to obtain the image frame file segmented by the image processing unit and output a detection image frame with a box selection mark;

[0043] An image comparison unit is used to obtain the detection image frame and retrieve the fish image database stored in the storage unit, and match the two and output the image matching result;

[0044] The judgment unit is used to judge the image matching result and generate an execution instruction on whether to output fish information.

[0045] As a preferred implementation option, preferably, the present solution also includes a visual output unit connected to the judgment unit, through which fish information can be intuitively output, and the video stream can be further synchronously displayed or displayed and output in the form of image frames.

[0046] Based on the above method scheme, the present scheme also provides a storage medium, in which at least one instruction, at least one program, code set or instruction set is stored, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the above-mentioned underwater video fish identification method.

[0047] By adopting the above-mentioned technical scheme, the present invention has the following beneficial effects compared with the prior art: the scheme cleverly constructs a fish image database to divide the underwater video images into image frames, and then performs image recognition to select the body parts of the fish in the video; by matching the framed marked area with the fish image database, it is possible to obtain better and faster image matching results while saving computing power as much as possible; by setting judgment conditions for the matching results to obtain matching results with higher confidence levels for output, the identified fish information is obtained; this method is not only reliable to implement and has high processing efficiency, but also has a good user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 A brief schematic diagram of the identification method of the present invention;

[0050] Figure 2 A simplified schematic diagram of a detection image frame with a box selection mark output during the processing of the recognition method of the present invention;

[0051] Figure 3 This is a schematic diagram of the data transmission process during the operation of the identification method of the present invention;

[0052] Figure 4 FIG. 1 is a simplified schematic diagram of one of the matching mechanisms of the identification method of the present invention, wherein only the schematic diagram of data transmission between the fish image database and the image comparison unit is shown, and the rest of the parts are similar to those of FIG. Figure 3Shown are substantially the same;

[0053] Figure 5 A schematic diagram of a simplified data determination process as one of the matching mechanisms of the identification method of the present invention, wherein the link part and Figure 3 Shown are substantially the same;

[0054] Figure 6 FIG. 1 is a simplified schematic diagram of the second matching mechanism of the recognition method of the present invention, in which only the data transmission between the image detection frame and the image comparison unit is changed, and the rest is the same as Figure 3 Shown are substantially the same;

[0055] Figure 7 This is one of the simplified schematic diagrams of the underwater video fish identification system of the present invention;

[0056] Figure 8 This is the second simplified schematic diagram of the underwater video fish identification system of the present invention. DETAILED DESCRIPTION

[0057] The present invention will be further described in detail below in conjunction with the accompanying drawings and examples. It is particularly noted that the following examples are only used to illustrate the present invention, but are not intended to limit the scope of the present invention. Similarly, the following examples are only partial embodiments of the present invention rather than all embodiments, and all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0058] like Figure 1 As shown, the present invention provides an underwater video fish identification method, comprising:

[0059] S01, build fish image database;

[0060] S02, obtaining underwater video images;

[0061] S03, dividing the underwater video image into image frame files, and then importing the files into the image recognition system for recognition processing, and outputting the detection image frames with box selection marks;

[0062] S04, extracting the area with the box selection mark in the detection image frame, importing it into the fish image database for matching processing, and outputting the image matching result;

[0063] S05. When the image matching result meets the preset conditions, the fish information is output.

[0064] Among them, reference Figure 2 As shown, as a possible implementation manner, further, the detection image frame with a box selection mark is a frame selection mark for the whole fish, the fish head, the fish body, the fish fins and the fish tail in the image frame;

[0065] The data stored in the fish image database includes multi-view overall image data of different fish and local image data of the fish head, body, fins and tail that are segmented and stored separately from multiple viewpoints.

[0066] As a preferred implementation option, preferably, the image matching result is also normalized to obtain a 1×2 matrix data result (x1, y1), where x1+y1=1, x1 is the matching probability, and y1 is the matching probability.

[0067] Combination Figure 3 As shown, as a preferred implementation option, preferably, the method of segmenting underwater video images into image frame files is:

[0068] The video stream in the underwater video image is converted into image frames, and then the corresponding image frames are extracted at preset time intervals to generate detection image frames, which are used to be imported into the image recognition system for recognition processing. When the probability of the image matching result is greater than the preset value, the fish information is output. If the traversal method is used and there is no match greater than the preset value in the fish database, the matching is terminated directly.

[0069] In terms of matching processing, different matching mechanisms have different efficiencies. Therefore, this solution also provides two data matching mechanisms, as follows:

[0070] exist Figures 1 to 3 Based on the above, further combining Figure 4 As one of the matching mechanisms, preferably, the fish data in the fish image database are also classified according to the body proportion characteristics and head characteristics of the fish and an index library is established. The body proportion characteristics are determined based on the length of the fish and the height ratio range from its abdomen to the back. The body proportion characteristic numerical range includes at least: [0, 1], [1, 2], [2, 5], [5, 10], [10, 15], [15, 30], [30, +∞]. Through the index library, the data in the fish image database can be classified and indexed, which can overcome the limitations of traditional traversal matching, such as: weak matching targeting and large data processing volume.

[0071] Further integration Figure 5As shown, in the matching mechanism of the present scheme, when the detection image frame is imported into the fish image database for matching, an edge detection algorithm is used to extract the overall outline of the fish in the detection image frame, estimate the height and length of the fish, and obtain the length-to-height ratio, which is then matched to the index library of the fish image database, and then matched with the data set under the corresponding body shape proportion feature value interval, and the fish data (i.e., temporary image matching result) with the highest probability of matching the fish in the detection image frame is obtained in real time, and the matching probability of the fish data is set as the first data;

[0072] When the first data is greater than a first preset value, it is outputted as a temporary image matching result first, and then when it meets a preset condition, the fish information is outputted;

[0073] When the first data is smaller than the second preset value, continue to call the data in the fish database for matching until the first data is larger than the second preset value or the traversal and matching of the fish database is completed;

[0074] Among them, after the matching of the data set corresponding to the body shape proportion feature value interval is completed, the data sets under the remaining body shape proportion feature value intervals are matched in order from small to large according to the difference between the starting value of the body shape proportion feature value interval and the length-to-height ratio of the fish in the current detection image frame.

[0075] As a preferred implementation option, preferably, the edge detection algorithm is a Canny edge detection algorithm or a Scharr edge detection algorithm.

[0076] exist Figures 1 to 3 Based on the reference Figure 5 As shown, further combined Figure 6 As the second matching mechanism, preferably, when the detection image frame is imported into the fish image database for matching, a detection neural network algorithm is also used to judge the area with a box selection mark in the detection image frame, and then the detection image frame is classified into a first detection set with fish head image information, a second detection set with fish body information but without fish head, and a third detection set without fish head and without fish body according to whether the fish head and fish body are reflected;

[0077] in,

[0078] The detection image in the first detection set uses a traversal method to match the area marked with fish heads in the detection image frame with the fish head data in the fish image database. When the matching probability is greater than a preset value, the area marked with the whole fish in the detection image frame is extracted for further matching with the multi-view overall image data of the fish corresponding to the fish head data, and the fish data with the highest probability of matching the fish in the detection image frame is obtained in real time, which is set as the second data;

[0079] When the second data is greater than the first preset value, it is output as a temporary image matching result first, and then the fish information is output when it meets the preset conditions.

[0080] When the second data is less than the second preset value, continue to call the data in the fish database for matching until the second data is greater than the second preset value or the traversal and matching of the fish database is completed;

[0081] The detection image in the second detection set uses a traversal method to match the area marked with the fish body in the detection image frame with the fish body data in the fish image database. When the matching probability is greater than a preset value, the area marked with the whole fish in the detection image frame is extracted for further matching with the multi-view overall image data of the fish corresponding to the fish body data, and the fish data with the highest probability of matching the fish in the detection image frame is obtained in real time, which is set as the third data;

[0082] When the third data is greater than the first preset value, it is outputted as a temporary image matching result first, and then the fish information is outputted when it meets the preset conditions.

[0083] When the third data is less than the second preset value, continue to call the data in the fish database for matching until the third data is greater than the second preset value or the traversal and matching of the fish database is completed;

[0084] The detection image in the third detection set uses a traversal method to match the area marked with the whole fish in the detection image frame with the multi-view whole image data of the fish in the fish image database, and obtains the fish data with the highest probability of matching the fish in the detection image frame in real time, which is set as the fourth data;

[0085] When the fourth data is greater than the first preset value, it is outputted as a temporary image matching result first, and then the fish information is outputted when it meets the preset conditions.

[0086] When the fourth data is smaller than the second preset value, the calling of the data in the fish database for matching continues until the fourth data is larger than the second preset value or the traversal and matching of the fish database is completed.

[0087] As a preferred implementation option, preferably, the fish image database also stores fish skin disease image data and fish abnormal posture image data, and the output image matching results include: fish information, skin disease information and abnormal state information.

[0088] As a preferred implementation option, preferably, when the matching result meets the preset conditions, the fish information is output in a visual manner;

[0089] Among them, the method of outputting fish information in a visual manner includes: drawing a marking line for the entire area of ​​the fish marked by the box selection in the detection image frame in real time, and outputting the fish information in text at the end of the marking line.

[0090] like Figure 7 As shown, based on the above method scheme, this scheme also provides an underwater video fish identification system, including:

[0091] A storage unit, used for storing a fish image database;

[0092] An image acquisition unit, used to acquire underwater video;

[0093] An image processing unit, used for segmenting underwater video into image frame files;

[0094] An image recognition unit is used to obtain the image frame file segmented by the image processing unit and output a detection image frame with a box selection mark;

[0095] An image comparison unit is used to obtain the detection image frame and retrieve the fish image database stored in the storage unit, and match the two and output the image matching result;

[0096] The judgment unit is used to judge the image matching result and generate an execution instruction on whether to output fish information.

[0097] Among them, Figure 8 As shown, as an extended solution, preferably, this solution also includes a visual output unit connected to the judgment unit, through which fish information can be output intuitively, and the video stream can be further displayed synchronously or displayed and output in the form of image frames.

[0098] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0099] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0100] The above descriptions are only some embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any equivalent device or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for identifying fish from underwater videos, characterized in that: include: Build fish image database; Acquire underwater video images; Segment the underwater video images into image frame files, then import them into the image recognition system for recognition processing, and output the detected image frames with box selection marks; Extract the area with the box mark in the detection image frame, import it into the fish image database for matching processing, and output the image matching result; When the image matching results meet the preset conditions, the fish information is output; The detection image frame with the frame selection mark is to mark the whole fish, the fish head, the fish body, the fish fins and the fish tail in the image frame respectively; The data stored in the fish image database includes multi-view overall image data of different fish and local image data of the fish head, body, fins and tail that are segmented and stored separately from multiple viewpoints; The fish data in the fish image database are further classified according to the body shape proportion characteristics and the head characteristics of the fish and an index library is established, wherein the body shape proportion characteristics are determined according to the length of the fish and the height ratio range from its abdomen to the back, and the body shape proportion characteristic value interval includes: [0, 1], [1, 2], [2, 5], [5, 10], [10, 15], [15, 30], [30, +∞]; When the detection image frame is imported into the fish image database for matching, an edge detection algorithm is used to extract the overall outline of the fish in the detection image frame, the height and length of the fish are estimated, and the length-to-height ratio is calculated, and the length-to-height ratio is matched to the index library of the fish image database, and then matching is performed in the data set under the corresponding body shape proportion feature value interval, and fish data with the highest probability of matching the fish in the detection image frame is obtained in real time, and the matching probability of the fish data is set as the first data; When the first data is greater than a first preset value, it is outputted as a temporary image matching result first, and then when it meets a preset condition, the fish information is outputted; When the first data is smaller than the second preset value, continue to call the data in the fish database for matching until the first data is larger than the second preset value or the traversal and matching of the fish database is completed; After the matching of the data set corresponding to the body shape proportion feature value interval is completed, the data sets corresponding to the body shape proportion feature value interval are matched in order from small to large according to the difference between the starting value of the body shape proportion feature value interval and the fish height ratio in the current detection image frame; When the detection image frame is imported into the fish image database for matching, a detection neural network algorithm is also used to judge the area with a box selection mark in the detection image frame, and then the detection image frame is classified into a first detection set with fish head image information, a second detection set with fish body information but without fish head, and a third detection set without fish head and without fish body according to whether the fish head and fish body are reflected; in, The detection image in the first detection set uses a traversal method to match the area marked with fish heads in the detection image frame with the fish head data in the fish image database. When the matching probability is greater than a preset value, the area marked with the whole fish in the detection image frame is extracted for further matching with the multi-view overall image data of the fish corresponding to the fish head data, and the fish data with the highest probability of matching the fish in the detection image frame is obtained in real time, which is set as the second data; When the second data is greater than the first preset value, it is output as a temporary image matching result first, and then the fish information is output when it meets the preset conditions. When the second data is less than the second preset value, continue to call the data in the fish database for matching until the second data is greater than the second preset value or the traversal and matching of the fish database is completed; The detection image in the second detection set uses a traversal method to match the area marked with the fish body in the detection image frame with the fish body data in the fish image database. When the matching probability is greater than a preset value, the area marked with the whole fish in the detection image frame is extracted for further matching with the multi-view overall image data of the fish corresponding to the fish body data, and the fish data with the highest probability of matching the fish in the detection image frame is obtained in real time, which is set as the third data; When the third data is greater than the first preset value, it is outputted as a temporary image matching result first, and then the fish information is outputted when it meets the preset conditions. When the third data is less than the second preset value, continue to call the data in the fish database for matching until the third data is greater than the second preset value or the traversal and matching of the fish database is completed; The detection image in the third detection set uses a traversal method to match the area marked with the whole fish in the detection image frame with the multi-view whole image data of the fish in the fish image database, and obtains the fish data with the highest probability of matching the fish in the detection image frame in real time, which is set as the fourth data; When the fourth data is greater than the first preset value, it is outputted as a temporary image matching result first, and then the fish information is outputted when it meets the preset conditions. When the fourth data is smaller than the second preset value, the calling of the data in the fish database for matching continues until the fourth data is larger than the second preset value or the traversal and matching of the fish database is completed.

2. The underwater video fish identification method according to claim 1, characterized in that: The image matching result is also normalized to obtain a 1×2 matrix data result (x1, y1), where x1+y1=1, x1 is the probability of matching, and y1 is the probability of not matching.

3. The underwater video fish identification method according to claim 2, characterized in that: The method of segmenting underwater video images into image frame files is: The video stream in the underwater video image is converted into image frames, and then the corresponding image frames are extracted at preset time intervals to generate detection image frames, which are used to be imported into the image recognition system for recognition processing.

4. The underwater video fish identification method according to claim 1, characterized in that: The edge detection algorithm is a canny edge detection algorithm or a Scharr edge detection algorithm.

5. The underwater video fish identification method according to claim 1 or 4, characterized in that: When the matching results meet the preset conditions, the fish information is output in a visual way; The method of visually outputting fish information includes: drawing a marking line for the entire fish area marked by the box in the detection image frame in real time, and outputting the fish information in text at the end of the marking line; The fish image database also stores fish skin disease image data and fish abnormal posture image data, and the output image matching result includes: fish information, skin disease information and abnormal state information.

6. An underwater video fish identification system, which is applied with the underwater video fish identification method according to any one of claims 1 to 5, characterized in that: include: A storage unit, used for storing a fish image database; An image acquisition unit, used to acquire underwater video; An image processing unit, used for segmenting underwater video into image frame files; An image recognition unit is used to obtain the image frame file segmented by the image processing unit and output a detection image frame with a box selection mark; An image comparison unit is used to obtain the detection image frame and retrieve the fish image database stored in the storage unit, and match the two and output the image matching result; The judgment unit is used to judge the image matching result and generate an execution instruction on whether to output fish information.

7. A storage medium, characterized in that: The storage medium stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the underwater video fish identification method as described in any one of claims 1 to 5.

Images