A method for detecting the activity of aquatic animals by fusing timing behavior and body feature

Abstract

The present application relates to the technical field of data caching, and more particularly to a method for detecting the activity of aquatic animals by fusing time sequence behavior and body state characteristics, comprising: determining whether to adjust the point position recognition analysis to parameter extraction analysis based on paragraph stability; when performing point position recognition analysis, determining the screening criterion to be abnormal point positions or frame-by-frame analysis based on body state active representation values, and determining whether the point position category is an abnormal point position based on point position stability and interference representation values; when performing parameter extraction analysis, determining curve confidence based on time sequence correlation and body state stability, and determining a reference curve based on the curve confidence; and determining whether to adjust the lowest limit based on capture coordination and capture stability. The present application can improve the effectiveness of collected images.

Description

Technical Field

[0001] This invention relates to the field of activity detection technology, and in particular to a method for detecting the activity of aquatic animals that integrates temporal behavior and physical characteristics. Background Technology

[0002] In aquaculture, automated activity detection based on video has been gradually applied. However, existing methods mostly adopt static strategies, such as setting fixed feed net locations, recording at fixed frame rates, and storing data indiscriminately. Since the temporal behavior and physical characteristics of shrimp groups change dynamically with feeding rhythms and the environment, fixed strategies are difficult to effectively represent the actual live state of shrimp groups, thereby reducing the recording effect of video frames and the reference value of stored data.

[0003] Chinese Patent Publication No. CN111914626B discloses a liveness detection / threshold adjustment method, apparatus, electronic device, and storage medium. The liveness threshold adjustment method includes the following steps: when a liveness detection operation is detected, obtaining the liveness confidence score obtained from the liveness detection operation; obtaining the current liveness threshold; comparing the liveness threshold with the liveness confidence score to obtain a comparison result; and adjusting the liveness threshold based on the comparison result. This application's embodiment dynamically adjusts the liveness threshold based on the result of the liveness detection operation, thereby reducing the probability of false positives and improving liveness detection efficiency to a certain extent. However, the above technical solution suffers from the following problem: it lacks consideration of external factors such as movement and occlusion during the acquisition process, which interfere with the liveness status and thus reduces the effectiveness of the acquired images. Summary of the Invention

[0004] Therefore, this invention provides a method for detecting the activity of aquatic animals that integrates temporal behavior and body shape characteristics, in order to overcome the problem in the prior art that does not take into account the interference of external factors such as movement and occlusion during the collection process on the live state, thus leading to a reduction in the effectiveness of the collected images.

[0005] To achieve the above objectives, the present invention provides a method for detecting the activity of aquatic animals by integrating temporal behavior and body posture characteristics, comprising: The decision to change from point identification analysis to parameter extraction analysis is based on the comparison between paragraph stability and preset paragraph stability. When performing point identification analysis, the selection criteria are determined as abnormal points or frame-by-frame analysis based on the comparison results between the body activity characterization value and the preset body activity characterization value; and the point category is determined as an abnormal point based on the point stability and interference characterization value. When performing parameter extraction analysis, the curve confidence level is determined based on temporal correlation and postural stability, and a reference curve is determined based on the comparison between the curve confidence level and the preset curve confidence level. The delay window length is determined based on the characteristics of the reference curve. Whether to adjust for the minimum limit is determined based on capture coordination and capture stability.

[0006] Furthermore, segment stability is determined based on the detection box variation coefficient and the inter-frame grayscale distribution difference coefficient; The stability of the segment is negatively correlated with the coefficient of variation of the detection box and the coefficient of difference in gray-scale distribution between frames.

[0007] Furthermore, for video frames with a segment stability greater than a preset segment stability, point identification and analysis are performed. For video frames whose segment stability is less than or equal to the preset segment stability, parameter extraction analysis is performed.

[0008] Furthermore, when performing location identification analysis, if the postural activity index is greater than the preset postural activity index, the location is determined to be an abnormal location based on the screening criteria. When the physical activity index is less than or equal to the preset physical activity index, the selection criterion is frame-by-frame analysis.

[0009] Furthermore, points whose point stability is greater than the preset point stability and whose interference characterization value is less than or equal to the preset interference characterization value are recorded as abnormal points.

[0010] Furthermore, when performing parameter extraction analysis, the confidence level of the curve is determined based on the temporal correlation degree and the body stability.

[0011] Furthermore, the extraction action where the curve confidence level is greater than the preset curve confidence level is recorded as the reference curve.

[0012] Furthermore, for the lifting state where the capture coordination degree is less than or equal to the preset capture coordination degree or the capture stability degree is greater than the preset capture stability degree, it is determined that the minimum limit of the material net should be adjusted based on the capture parameters; The capture parameters include the live organism escape rate and the residual rate in the effluent.

[0013] Furthermore, during the adjustment of the minimum limit of the feeder, if the live animal jump rate is greater than the preset live animal jump rate, it is determined that the minimum adjustment limit should be reduced based on the first effective characterization value. If the residual rate of effluent is greater than the preset residual rate of effluent, the minimum adjustment limit will be increased based on the second effective characterization value. The first effective characterization value is determined based on the live body jump rate and the preset live body jump rate; The second effective characterization value is determined based on the effluent residual rate and the preset effluent residual rate.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: In the technical solution of the present invention, the stability of a segment is determined based on the detection box variation coefficient and the inter-frame gray-scale distribution difference coefficient, and the segment stability is negatively correlated with the detection box variation coefficient and the inter-frame gray-scale distribution difference coefficient, respectively. The detection box variation coefficient characterizes the degree of fluctuation in the number of live targets during the data monitoring period, and the inter-frame gray-scale distribution difference coefficient characterizes whether the video frame content is continuously changing. This avoids the one-sidedness of stability assessment caused by relying on only a single fluctuation index or ignoring the temporal changes in gray-scale distribution in the prior art. It is difficult to adapt to the actual working conditions of simultaneous sudden changes in the number of live targets and continuous changes in the picture content in the monitoring scenario, which affects the reasonable selection and processing efficiency of subsequent segment analysis methods. In this way, the comprehensiveness and accuracy of video segment stability assessment are improved, providing a reliable quality basis for segment segmentation, key frame extraction or anomaly detection in dynamic scenes.

[0015] In this invention, for video frames with a segment stability greater than a preset segment stability, point identification analysis is performed; for video frames with a segment stability less than or equal to the preset segment stability, parameter extraction analysis is performed. By setting a segment stability threshold for differentiated processing, the invention avoids the waste of computing resources or loss of key information caused by using a uniform analysis strategy for all video frames in the prior art. That is, when the segment stability is high, point identification can be performed directly to quickly obtain the target location, while when the segment stability is low, parameter extraction analysis is performed to capture dynamic features. This significantly improves the comprehensiveness and accuracy of target detection and behavior recognition in complex monitoring scenarios while ensuring the real-time performance of the analysis, and enhances the adaptive ability of the video analysis system to dynamic changes in the scene.

[0016] In the technical solution of this invention, when performing point recognition analysis, if the body activity characterization value is greater than the preset body activity characterization value, the selection criterion is determined to be an abnormal point; if the body activity characterization value is less than or equal to the preset body activity characterization value, the selection criterion is determined to be frame-by-frame analysis. By dynamically selecting the appropriate selection criterion through body activity characterization value, the computational redundancy or key anomaly omission problems caused by using fixed points or full-frame scanning in the prior art are avoided. That is, when the target body activity is high, abnormal points are located first to achieve rapid response; when the body activity is low, frame-by-frame fine analysis is used to capture subtle changes. Thus, while ensuring efficient processing of high-activity scenes, the sensitivity and completeness of abnormal behavior detection in low-activity scenes are improved, and the adaptive ability and recognition robustness of the video behavior analysis system to different body activity levels are enhanced.

[0017] In this invention, for fishing net lifting states where the capture coordination degree is less than or equal to a preset capture coordination degree or the capture stability degree is greater than a preset capture stability degree, the minimum limit of the net is adjusted based on capture parameters. The capture parameters include the live fish escape rate and the residual rate in the effluent. The minimum limit adjustment mechanism is triggered by the dual conditions of capture coordination degree and capture stability degree. The live fish escape rate represents the degree of target escape and the residual rate in the effluent represents the separation efficiency. This avoids the problem of insufficient adaptability caused by using fixed net limits or relying on a single working condition in the prior art. It is difficult to balance capture rate and sorting quality when the fishing environment changes dynamically. Thus, the minimum limit of the net is dynamically optimized under working conditions with poor capture coordination degree or excessive stability, effectively reducing the live fish escape rate and the residual rate in the effluent. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the aquatic animal activity detection method that integrates temporal behavior and body shape characteristics according to the present invention; Figure 2 This is a flowchart illustrating how the analysis method is determined based on paragraph stability in this invention. Figure 3 This is a flowchart illustrating the process of determining screening criteria based on physical activity characterization values ​​in this invention. Figure 4 This is a flowchart illustrating the process of determining whether to adjust the minimum limit of the feeder based on capture coordination and capture stability, according to the present invention. Detailed Implementation

[0019] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0020] Please see Figures 1 to 4 As shown, this invention provides a method for detecting the activity of aquatic animals that integrates temporal behavior and body posture characteristics, comprising: The decision to change from point identification analysis to parameter extraction analysis is based on the comparison between paragraph stability and preset paragraph stability. When performing point identification analysis, the selection criteria are determined as abnormal points or frame-by-frame analysis based on the comparison results between the body activity characterization value and the preset body activity characterization value; and the point category is determined as an abnormal point based on the point stability and interference characterization value. When performing parameter extraction analysis, the curve confidence level is determined based on temporal correlation and postural stability, and a reference curve is determined based on the comparison between the curve confidence level and the preset curve confidence level. The delay window length is determined based on the characteristics of the reference curve. Whether to adjust for the minimum limit is determined based on capture coordination and capture stability.

[0021] The application scenario of this invention is to detect the condition of shrimp in a breeding area. In this invention, the shrimp that need to be detected are recorded as target live organisms. A feed net that can move up and down is set up, and a high-definition camera is set on the feed net to record video. The recorded video is recorded as the video segment to be analyzed.

[0022] Specifically, segment stability is determined based on the detection box variation coefficient and the inter-frame grayscale distribution difference coefficient; The stability of the segment is negatively correlated with the coefficient of variation of the detection box and the coefficient of difference in gray-scale distribution between frames.

[0023] Obtain the video segment to be analyzed within the most recent data monitoring period. For each video segment, uniformly set several points within the segment and record each point as a detection time. Obtain the number of target detection boxes at each detection time. The coefficient of variation of the detection boxes is calculated as follows: The Let be the standard deviation of the number of object detection boxes in the video segment to be analyzed. This represents the average number of target detection boxes in the video segment to be analyzed.

[0024] The method for confirming the inter-frame grayscale distribution difference coefficient is as follows: Inter-frame grayscale distribution difference coefficient = Duration of the marked segment / Duration of the video segment to be analyzed; extract the video frames at each detection time, and convert each video frame into a grayscale image, for the first... Frame video frame and the first Chi-square distance of video frames = ;in, =0,1,…, , = / N, where N is the nth Total number of pixels within a video frame. For the first The grayscale value in a video frame is The number of pixels, The grayscale level is denoted by .

[0025] = Bit depth characterizes video frame resolution; a higher bit depth results in finer grayscale resolution and a greater ability to distinguish brightness levels. Users can adaptively set the bit depth value according to their specific application needs. The higher the user's requirements for video frame analysis, the larger the bit depth value. In this invention, the bit depth is set to 8, with each pixel using 0... 255 represents brightness, with a total of 256 gray levels. =256. If >Activity threshold, then mark the first The video frame is the active frame, and the longest consecutive duration of the active frame is recorded as the marked segment. In this invention, the length of the data monitoring period is 7 days.

[0026] Segment stability = Detection box variation coefficient / Detection box variation threshold + Inter-frame gray-scale distribution difference coefficient / Inter-frame gray-scale distribution difference threshold; In this invention, the detection box variation threshold is 0.5, and the inter-frame gray-scale distribution difference threshold is 0.2. It is easy to understand that the detection box variation coefficient represents the degree of fluctuation in the number of live targets during the data monitoring period, and the inter-frame gray-scale distribution difference coefficient represents whether the video frame content is continuously changing. The segment stability is determined based on the detection box variation coefficient and the inter-frame gray-scale distribution difference coefficient, which is used to determine the subsequent segment analysis method. The higher the user's tolerance for the impact of the detection box variation coefficient on segment stability, the larger the value of the detection box variation threshold; the higher the user's tolerance for the impact of the inter-frame gray-scale distribution difference coefficient on segment stability, the larger the value of the inter-frame gray-scale distribution difference threshold.

[0027] Specifically, for video frames with a segment stability greater than the preset segment stability, point identification and analysis are performed. For video frames whose segment stability is less than or equal to the preset segment stability, parameter extraction analysis is performed.

[0028] In this invention, the preset segment stability value is 1.5. It can be understood that the higher the user's requirements for the activity detection effect of the target liveness, the smaller the preset segment stability value. When the segment stability is greater than the preset segment stability, it indicates that at least one of the detection box variation coefficient or the inter-frame grayscale distribution difference coefficient is at a high level. At this time, there is obvious fluctuation in the number of liveness objects or continuous change in the image content within the segment, making it suitable for point-based identification analysis to quickly capture abnormal points. When the segment stability is less than or equal to the preset segment stability, it indicates that the segment is relatively stable overall, with small fluctuations in the number of liveness objects and the image content, making it suitable for parameter extraction analysis to obtain detailed behavioral curve features.

[0029] The higher the user's requirement for the stability of liveness detection, the smaller the preset segment stability value should be, so that more segments are judged as unstable and transferred to parameter extraction analysis; conversely, the higher the requirement for real-time detection, the larger the preset segment stability value should be, so that more segments are judged as stable and point identification analysis is performed. In this invention, 1.5 is selected as the optimal value based on the balance between detection accuracy and computational efficiency, verified in a classic aquaculture scenario.

[0030] Specifically, when performing location identification and analysis, if the value of body activity is greater than the preset value of body activity, the location is determined to be an abnormal location based on the screening criteria. When the physical activity index is less than or equal to the preset physical activity index, the selection criterion is frame-by-frame analysis.

[0031] For a single video segment to be analyzed, target live bodies whose posture representation value within the video frame is greater than the preset posture representation value are marked live bodies. For a single target live body, the posture representation value = projected length / actual body length. The bounce reference degree = number of marked live bodies / number of target live bodies within the video frame. Video frames whose bounce reference degree is greater than the preset bounce reference degree are marked as reference video frames. The body activity representation value = number of reference video frames / total number of video frames.

[0032] The activity level of the target live body within the video segment to be analyzed is represented by the body activity characterization value. If the target live body is relatively active as a whole, only abnormal points are screened. If the target live body is relatively inactive as a whole, all video frames are analyzed one by one, that is, there is no need to screen video frames.

[0033] In this invention, the preset value of the body activity characterization value is 0.7. It can be understood that the body activity characterization value represents the overall activity level of the target living body in the video segment to be analyzed, and the video frame to be screened is determined based on the overall activity level of the video segment to be analyzed. If the overall activity level is high, only the abnormal points are analyzed.

[0034] Specifically, points whose stability is greater than the preset point stability and whose interference characterization value is less than or equal to the preset interference characterization value are recorded as abnormal points.

[0035] For a single detection moment, the reciprocal of the cumulative Euclidean distance between the center point of the detection frame corresponding to that detection moment and adjacent frames is recorded as the position offset; the normalized value of the average rate of change of the width and height of the detection frame between adjacent frames is recorded as the size change rate, and the point stability = 1 / (position offset × α + size change rate × β); where α and β are weighting coefficients, which can be adaptively set by the user according to the scene stability requirements. The higher the scene stability requirements, the larger the values ​​of α and β. The greater the user's tolerance for the influence of position offset on point stability, the larger the value of α. The greater the user's tolerance for the influence of size change rate on point stability, the larger the value of β. In this invention, α = 0.5 and β = 0.5.

[0036] In this invention, the preset point stability value is 0.6. It can be understood that the point stability characterizes the degree of fluctuation of the spatial position and apparent size of the target living body in several consecutive frames. The point stability is compared with the preset point stability to determine whether the point is a normal point. If the point stability is greater than the preset point stability, the point is determined to have the basic conditions for further analysis. Otherwise, the point is determined to have too large a fluctuation and is excluded from the screening range of abnormal points.

[0037] In this invention, the preset interference characterization value is set to 0.3. This means that the interference characterization value represents the degree of interference of external factors such as occlusion, sudden changes in lighting, or water ripples in the video frame on the detection of liveness of the target. The interference characterization value is compared with the preset interference characterization value to determine whether the interference can be ignored. If the interference characterization value is less than or equal to the preset interference characterization value, the interference is considered to be within an acceptable range. At this time, the stability of the point is combined to further determine whether it is an abnormal point. Otherwise, the interference is considered to be too large, and the point has no analytical value and will not be included in the abnormal point identification process.

[0038] Specifically, when performing parameter extraction analysis, the confidence level of the curve is determined based on the temporal correlation degree and the stability of the body.

[0039] The curve confidence score is used to characterize the reliability of the motion curve extracted based on temporal correlation and body stability. When performing extraction parameter analysis, for the continuous motion sequence in the video segment to be analyzed, the curve confidence score is determined based on temporal correlation and body stability. The temporal correlation reflects the coherence of the target's posture changes between adjacent frames, and the body stability reflects the degree of deformation fluctuation of the target's contour within the time period covered by the motion curve. The curve confidence score is positively correlated with both the temporal correlation and the body stability.

[0040] The confidence level of the curve is calculated as follows: time correlation degree × γ + body stability degree × δ, where γ and δ are weighting coefficients, and γ + δ = 1; in this invention, γ is 0.5 and δ is 0.5.

[0041] Specifically, the extraction action where the confidence level of the curve is greater than the preset confidence level of the curve is recorded as the reference curve.

[0042] In this invention, the preset curve confidence level is set to 0.75. This preset curve confidence level serves as a critical value for determining whether an extracted action can be used as a reference curve. If the curve confidence level is greater than the preset curve confidence level, the extracted action is deemed to have sufficient reliability and is recorded as a reference curve for determining the subsequent delay window length. If the curve confidence level is less than or equal to the preset curve confidence level, the extracted action is deemed insufficiently reliable and is not used as a reference curve. The higher the accuracy requirement for action analysis, the larger the preset curve confidence level should be.

[0043] Specifically, the length of the delay window is determined based on the duration of the reference curve and the rate of change of curvature.

[0044] In this invention, the reference curve is the continuous trajectory of the extraction action where the curve confidence is greater than a preset curve confidence. The delay window length is determined based on the duration of the reference curve and the rate of curvature change; the delay window length is used to limit the time range for subsequent frame extraction or action prediction.

[0045] The duration of the reference curve is the total time elapsed from the start frame to the end frame. Specifically, the duration is recorded as the time between the start and end times of the action corresponding to the reference curve. The longer the duration, the more substantial the temporal span of the action corresponding to the reference curve, and the higher its reference value for subsequent analysis.

[0046] The curvature change rate is used to characterize the degree of change in the curvature of the reference curve. For each trajectory point on the reference curve, the first and second derivatives at that point are calculated to determine the curvature at that point; the curvature change rate = effective difference / total number of frames of the reference curve, where the effective difference is the sum of the absolute values ​​of the curvature differences between all adjacent trajectory points within the reference curve. The larger the curvature change rate, the more complex the deformation of the reference curve and the richer the detailed features of the motion.

[0047] The method for determining the delay window length based on the duration and rate of curvature change of the reference curve is as follows: Delay window length = Duration / Preset duration + Rate of curvature change / Preset rate of curvature change. It can be understood that the duration reflects the temporal span of the reference curve, and the rate of curvature change reflects the morphological complexity of the reference curve. When the duration is long, the motion tends to be stable, and the delay window needs to be appropriately extended to capture subsequent margins. When the rate of curvature change is large, the motion details are rich, and the delay window needs to be appropriately shortened to avoid introducing irrelevant noise. Through the above weighted calculation, a reasonable delay window length can be adaptively determined, avoiding under-capture or over-capture problems caused by using a fixed delay window.

[0048] Specifically, for a lifting state where the capture coordination degree is less than or equal to the preset capture coordination degree or the capture stability degree is greater than the preset capture stability degree, the minimum limit of the material net is adjusted based on the capture parameters. The capture parameters include the live organism escape rate and the residual rate in the effluent.

[0049] Capture Coordination = 1 - The average number of captures is denoted as the capture stability.

[0050] Specifically, during the adjustment of the minimum limit of the feeder, if the live animal jump rate is greater than the preset live animal jump rate, the minimum limit of adjustment is reduced based on the first effective characterization value. If the residual rate of effluent is greater than the preset residual rate of effluent, the minimum adjustment limit will be increased based on the second effective characterization value. The first effective characterization value is determined based on the live body jump rate and the preset live body jump rate; The second effective characterization value is determined based on the effluent residual rate and the preset effluent residual rate.

[0051] First effective characterization value = live body jump rate / preset live body jump rate; second effective characterization value = residual effluent rate / preset residual effluent rate; when adjusting the minimum limit based on the first effective characterization value, the minimum limit = reference height × (first effective characterization value - 1); when adjusting the minimum limit based on the second effective characterization value, the minimum limit = reference height × (second effective characterization value + 1); the reference height is the minimum distance between the center point of the bottom of the feed tray and the bottom of the pool.

[0052] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for detecting the activity of aquatic animals by integrating temporal behavior and body posture characteristics, characterized in that, include: The decision to change from point identification analysis to parameter extraction analysis is based on the comparison between paragraph stability and preset paragraph stability. When performing point identification analysis, the selection criteria are determined as abnormal points or frame-by-frame analysis based on the comparison results between the body activity characterization value and the preset body activity characterization value; and the point category is determined as an abnormal point based on the point stability and interference characterization value. When performing parameter extraction analysis, the curve confidence level is determined based on temporal correlation and postural stability, and a reference curve is determined based on the comparison between the curve confidence level and the preset curve confidence level. The delay window length is determined based on the characteristics of the reference curve. Whether to adjust for the minimum limit is determined based on capture coordination and capture stability.

2. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 1, characterized in that, Segment stability is determined based on the detection box variation coefficient and the inter-frame grayscale distribution difference coefficient. The stability of the segment is negatively correlated with the coefficient of variation of the detection box and the coefficient of difference in gray-scale distribution between frames.

3. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 2, characterized in that, For video frames with a segment stability greater than the preset segment stability, determine and perform point identification analysis; For video frames whose segment stability is less than or equal to the preset segment stability, parameter extraction analysis is performed.

4. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 3, characterized in that, When performing location identification and analysis, if the postural activity index is greater than the preset postural activity index, the location is determined to be an abnormal location based on the screening criteria. When the physical activity index is less than or equal to the preset physical activity index, the selection criterion is frame-by-frame analysis.

5. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 4, characterized in that, Points with a stability greater than the preset point stability and an interference characterization value less than or equal to the preset interference characterization value are recorded as abnormal points.

6. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 5, characterized in that, When performing parameter extraction analysis, the confidence level of the curve is determined based on the temporal correlation degree and the stability of the body.

7. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 6, characterized in that, The extraction action of a curve with a confidence level greater than the preset curve confidence level is recorded as the reference curve.

8. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 7, characterized in that, The length of the delay window is determined based on the duration of the reference curve and the rate of change of curvature.

9. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 2, characterized in that, For fishing net lifting states where the capture coordination degree is less than or equal to the preset capture coordination degree or the capture stability degree is greater than the preset capture stability degree, the minimum limit of the fishing net is adjusted based on the capture parameters. The capture parameters include the live organism escape rate and the residual rate in the effluent.

10. The method for detecting the activity of aquatic animals by integrating temporal behavior and body shape characteristics according to claim 9, characterized in that, During the adjustment of the minimum limit of the feeder, if the live animal jump rate is greater than the preset live animal jump rate, the minimum limit of adjustment is reduced based on the first effective characterization value. If the residual rate of effluent is greater than the preset residual rate of effluent, the minimum adjustment limit will be increased based on the second effective characterization value. The first effective characterization value is determined based on the live body jump rate and the preset live body jump rate; The second effective characterization value is determined based on the effluent residual rate and the preset effluent residual rate.

Images