Target anomaly detection method and device applied to intelligent grain depot

Abstract

Embodiments of the present disclosure disclose a target anomaly detection method and device applied to an intelligent grain depot. A specific implementation of the method comprises: capturing a first detection image group sequence corresponding to a first detection device assembly through a first shooting device; capturing a second detection image group sequence corresponding to a second detection device assembly through a second shooting device; performing grain pile anomaly detection according to the first detection image group sequence and the second detection image group sequence to obtain a middle layer detection information set and a bottom layer detection information set of the grain pile; generating a three-dimensional statistical diagram of the grain depot; performing surface layer path planning of the grain pile based on the three-dimensional statistical diagram of the grain depot to obtain a surface layer detection path of the grain pile; performing fixed-point sampling of a grain surface to obtain a grain surface sampling image group; and generating a grain depot anomaly detection result according to the grain surface sampling image group, the three-dimensional statistical diagram of the grain depot and preset grain depot storage information. The implementation can improve the accuracy of the grain depot anomaly detection result.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the fields of grain storage detection technology, image recognition technology, path planning technology, and computer technology, and specifically to a target anomaly detection method and apparatus applied to intelligent grain depots. Background Technology

[0002] Anomaly detection can be used to determine if there are any abnormalities in a grain warehouse. Currently, when detecting anomalies in smart grain warehouses, taking pest detection as an example, the tendency of grain insects to climb and move upwards is often used to place insect traps on the surface or shallow layer of the grain pile to capture pests in the grain warehouse, thereby determining the severity of the pest infestation.

[0003] However, when using the above methods for target anomaly detection, the following technical problems often arise:

[0004] Different grain varieties are susceptible to different pests, and the habits of different pests vary, resulting in different distribution depths of pests within the grain pile. Surface sampling methods are insufficient for accurate pest sampling, making it difficult to determine the actual distribution and severity of pests. Consequently, abnormal detection results may contain significant discrepancies. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure propose a target anomaly detection method and apparatus for use in intelligent grain depots to address the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide a target anomaly detection method applied to a smart grain depot. The method includes: retrieving a first detection device component and capturing a first detection image sequence corresponding to the first detection device component using a first imaging device, wherein the first detection device component includes at least one first detection device, which is an insect trapping device installed inside the grain pile; retrieving a second detection device component and capturing a second detection image sequence corresponding to the second detection device component using a second imaging device, wherein the second detection device component includes at least one second detection device, which is an insect trapping device installed at the ventilation opening of the trough. The first detection device and... The second detection device is paired with the first detection image group sequence and the second detection image group sequence to perform grain pile anomaly detection, obtaining a grain pile middle layer detection information set and a grain pile bottom layer detection information set; in response to determining that the grain pile middle layer detection information set and the grain pile bottom layer detection information set meet the preset anomaly conditions, a three-dimensional statistical map of the grain warehouse is generated; based on the grain warehouse three-dimensional statistical map, a grain pile surface path is planned to obtain a grain pile surface detection path; based on the grain pile surface detection path and the intelligent grain turning machine, fixed-point sampling of the grain surface is performed to obtain a grain surface sampling image group; based on the grain surface sampling image group, the grain warehouse three-dimensional statistical map, and the preset grain warehouse storage information, a grain warehouse anomaly detection result is generated.

[0008] Secondly, some embodiments of this disclosure provide a target anomaly detection device for intelligent grain depots. The device includes: a first retrieval and imaging unit configured to retrieve a first detection device component and capture a first detection image sequence corresponding to the first detection device component using a first imaging device, wherein the first detection device component includes at least one first detection device, which is an insect trapping device laid inside the grain pile; and a second retrieval and imaging unit configured to retrieve a second detection device component and capture a second detection image sequence corresponding to the second detection device component using a second imaging device, wherein the second detection device component includes at least one second detection device, which is an insect trapping device installed at the ventilation opening of the trough, and the first detection device and the second detection device correspond one-to-one; the grain pile anomaly detection... The detection unit is configured to perform grain pile anomaly detection based on the first detection image group sequence and the second detection image group sequence, and obtain a grain pile middle layer detection information set and a grain pile bottom layer detection information set; the first generation unit is configured to generate a grain warehouse three-dimensional statistical map in response to determining that the grain pile middle layer detection information set and the grain pile bottom layer detection information set meet preset anomaly conditions; the detection path planning unit is configured to perform grain pile surface path planning based on the grain warehouse three-dimensional statistical map, and obtain a grain pile surface detection path; the grain surface fixed-point sampling unit is configured to perform grain surface fixed-point sampling based on the grain pile surface detection path and the intelligent grain turning machine, and obtain a grain surface sampling image group; the second generation unit is configured to generate a grain warehouse anomaly detection result based on the grain surface sampling image group, the grain warehouse three-dimensional statistical map and preset grain warehouse storage information.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0011] The above-described embodiments of this disclosure have the following beneficial effects: the target anomaly detection method applied to intelligent grain depots through some embodiments of this disclosure can improve the accuracy of grain depot anomaly detection. Specifically, the reason for the large deviation in grain depot anomaly detection results is that different grain types correspond to different pests, and the habits of different pests are not entirely the same, resulting in different distribution depths of pests within the grain pile. Surface sampling methods are difficult to accurately sample pests, making it difficult to determine the actual distribution and severity of pests, thus leading to a large deviation in the anomaly detection results. Based on this, the target anomaly detection method applied to intelligent grain depots through some embodiments of this disclosure firstly retrieves a first detection device component and captures a first detection image sequence corresponding to the first detection device component using a first imaging device. The first detection device component includes at least one first detection device, which is an insect-trapping device laid inside the grain pile. Here, by introducing the first detection device component, it can be used to comprehensively detect pest data inside the grain pile. Simultaneously, the second detection equipment component is retrieved, and a second detection image sequence corresponding to the second detection equipment component is captured by the second imaging device. The second detection equipment component includes at least one second detection device, which is an insect-catching device installed at the ventilation opening of the grain trough. The first detection device corresponds one-to-one with the second detection device. Here, the second detection equipment component can be used to detect pest data at the bottom of the grain pile. Then, based on the aforementioned first and second detection image sequence sequences, grain pile anomaly detection is performed to obtain a middle layer detection information set and a bottom layer detection information set. This can be used to summarize pest information in the middle and bottom layers of the grain pile. Next, in response to determining that the aforementioned middle layer and bottom layer detection information sets meet preset anomaly conditions, a three-dimensional statistical map of the grain warehouse is generated. This can be used to determine the pest distribution within the grain warehouse. Then, based on the aforementioned three-dimensional statistical map of the grain warehouse, a surface path is planned for the grain pile to obtain a surface detection path. Afterwards, based on the aforementioned surface detection path and the intelligent grain turning machine, fixed-point sampling of the grain surface is performed to obtain a grain surface sampling image set. Here, by using surface path planning and fixed-point sampling of the grain pile, not only can supplementary detection be conducted to determine pest information on the grain surface, but it also further improves the monitoring of pest distribution, allowing for more precise control over the severity of pests inside the grain pile. Finally, based on the aforementioned grain surface sampling image set, the aforementioned 3D statistical map of the grain silo, and the preset grain storage information, grain silo anomaly detection results are generated. This significantly improves the accuracy of grain silo anomaly detection results. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a flowchart of some embodiments of the target anomaly detection method applied to intelligent grain depots according to the present disclosure;

[0014] Figure 2 This is a schematic diagram showing the distribution of the first testing equipment;

[0015] Figure 3 This is a schematic diagram showing the distribution of the second inspection equipment;

[0016] Figure 4 This is a schematic diagram of the surface path planning of the grain pile;

[0017] Figure 5 This is a structural schematic diagram of some embodiments of a target anomaly detection device applied to a smart grain depot according to the present disclosure;

[0018] Figure 6 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0019] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0020] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0021] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0022] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0023] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0024] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0025] Figure 1 A flow 100 of some embodiments of a target anomaly detection method for smart grain depots according to the present disclosure is shown. The target anomaly detection method for smart grain depots includes the following steps:

[0026] Step 101: Recover the first detection device component and capture the first detection image group sequence corresponding to the first detection device component using the first imaging device.

[0027] In some embodiments, the execution entity (e.g., a computing device) of the target anomaly detection method applied to a smart grain depot can retrieve the first detection device component via a wired or wireless means, and capture a first detection image sequence corresponding to the first detection device component via a first imaging device. The first detection device component includes at least one first detection device, which is an insect trapping device laid inside the grain pile.

[0028] As an example, see Figure 2 The diagram shows the distribution of the first detection equipment. Figure 2 In the flat-roofed grain warehouse shown, the surface where the red line is located is the grain surface, and the area below the surface where the red line is located is the grain pile. The first detection devices 201 in the first detection equipment assembly are evenly arranged inside the grain pile to capture pests generated inside the grain pile.

[0029] It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G / 5G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (Ultra Wide Band) connections, and other currently known or future wireless connection methods.

[0030] It should be noted that the aforementioned computing devices can be either hardware or software. When the computing device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device. When the computing device is software, it can be installed on the hardware devices listed above. It can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are made here.

[0031] In some optional implementations of certain embodiments, the first detection device in the first detection device assembly is fixedly connected to the first end of the electric telescopic rod, and the second end of the electric telescopic rod is fixedly connected to the top of the grain silo. Here, the first end can be the telescopic head of the electric telescopic rod, and the second end can be the fixed end of the electric telescopic rod.

[0032] The aforementioned executing entity recovers the first detection equipment component and captures a first detection image sequence corresponding to the first detection equipment component using the first imaging device, including:

[0033] For each of the first detection devices in the aforementioned first detection device assembly, the following steps are performed:

[0034] Step S1: Control the electric telescopic pole corresponding to the first detection device to retrieve the first detection device, and control the automated guided vehicle to retrieve the captured equipment from the first detection device and return it to the first shooting area. The automated guided vehicle is equipped with a cargo pallet. Next, by controlling the retraction of the electric telescopic pole, the corresponding first detection device can be removed from the grain pile.

[0035] In practice, within a flat-roofed grain silo, the aforementioned implementing entity can control a tracked automated guided vehicle (AGV) to move across the grain surface to the area directly beneath the first detection device. Then, it can control an electric telescopic boom to extend the first detection device to a preset height, after which the device opens its lower opening to allow the captured material to be dumped onto the cargo pallet of the intelligent transport vehicle. Finally, the AAV returns to the first imaging area, allowing the first imaging device to photograph the cargo pallet on the AAV, obtaining the first set of detection images. Furthermore, before moving to receive the next captured material from the first detection device, the AAV can be moved to a recovery area, where a lifting device on the AAV clears the captured material from the cargo pallet. This avoids interaction and conflict between captured materials from different detection devices. Here, a lifting device is fixed to the AAV, with its other end connected to the bottom of the cargo pallet via a rotating shaft.

[0036] Step S2: The first imaging device captures images of the first imaging area (i.e., the cargo pallet of the automated guided vehicle) to obtain a first detection image group. Each first detection image in the first detection image group is a consecutive frame image. The first imaging device can be a high-definition camera.

[0037] Step 102: Recover the second detection device component and capture the second detection image group sequence corresponding to the second detection device component using the second imaging device.

[0038] In some embodiments, the execution entity may retrieve the second detection device component and capture a second detection image sequence corresponding to the second detection device component using a second imaging device. The second detection device component includes at least one second detection device, which is an insect-catching device installed at the ventilation opening of the trench. The first detection device corresponds one-to-one with the second detection device.

[0039] As an example, see Figure 3 The diagram shows the distribution of the second inspection equipment. Figure 3 As shown, a concave ventilation trough 301 is provided at the bottom of the grain silo (only the location of the ventilation trough in the grain silo is shown in the figure). Multiple ventilation openings 302 are evenly distributed on the ventilation trough 301. Furthermore, the ventilation openings are equipped with mesh panels, allowing a second detection device to be installed directly below the ventilation openings (i.e., the mesh panels) for pest capture in the bottom layer of the grain pile.

[0040] In some optional implementations of certain embodiments, the second detection device in the second detection equipment assembly is fixed on a tracked remote-controlled transport vehicle. The track of the remote-controlled transport vehicle is laid in the ventilation trough of the grain silo. Each second detection device in the second detection equipment assembly has a corresponding position number. Here, the remote-controlled transport vehicle may be equipped with a positioning device to determine the coordinates during the transportation of the second detection device.

[0041] The aforementioned executing entity recovers the second detection equipment component and captures a second detection image sequence corresponding to the second detection equipment component using the second imaging device, including:

[0042] According to the location number corresponding to the second testing device, perform the following steps for each second testing device included in the second testing device component:

[0043] Step S1 involves controlling the corresponding remote-controlled transport vehicle to collect the second detection device and place the captured items from the second detection device into the second imaging area. Since only one track is set within each ventilation trough, and each second detection device within each trough has a corresponding position queue, with each position queue including its position number, multiple remote-controlled transport vehicles must travel sequentially. Therefore, the second detection device with the corresponding position number can be selected and collected in reverse order of the position queue. Here, the corresponding remote-controlled transport vehicle can be remotely moved out of the ventilation trough along the track. Then, the captured items from the second detection device are dumped into the second imaging area using manual or robotic arm control. Next, the second imaging device is controlled to capture images of the second imaging area, obtaining a second set of detection images. In practice, after transporting the second detection device to the designated trough ventilation opening, the remote-controlled transport vehicle can also use a lifting device to push the second detection device into close contact with the mesh panel of the trough ventilation opening to facilitate pest capture.

[0044] Step S2: The second imaging area is captured by the second imaging device to obtain the second detection image group. The second imaging device is a high-definition camera with a fixed imaging area.

[0045] Here, the positions of the first and second detection devices can be in a one-to-one correspondence. That is, the first detection device is positioned directly above the second detection device. Therefore, the first and corresponding second detection devices can perform simultaneous detection and recovery, facilitating the simultaneous determination of multi-layered pest information in a specific area within the grain pile.

[0046] In practice, the first and second detection devices can be the same or different insect-trapping devices. For example, based on the characteristics of different grain varieties and the characteristics of the corresponding pests, the first and second detection devices can be set to be the same insect-trapping device. For instance, both the first and second detection devices can be probe-type insect-trapping devices. Alternatively, based on the habits of different pests at different structural levels within the grain pile, the first and second detection devices can be different insect-trapping devices. For example, the first detection device can be a probe-type insect-trapping device, and the second detection device can be a bait-type insect trap.

[0047] For example, the characteristics of different grain types and their corresponding pests can be as follows: If the grain is wheat, corn, or other cereals, its large grains are prone to being bored, so the upper layer of the grain pile is prone to rice weevils, the middle layer is prone to wood-boring insects, and the lower layer is prone to mites and grain borers. If the grain is legumes, the upper layer of the grain pile is prone to adult green bean weevils and pea weevils, the middle layer is prone to bean weevil larvae and mixed flour beetles, and the lower layer is prone to mites. If the grain pile is rice, the upper layer of the grain pile is prone to adult rice weevils and wheat moths, the middle layer is prone to rice weevil larvae, and the lower layer is prone to dust mites and red flour beetles.

[0048] Step 103: Perform anomaly detection on the grain pile based on the first detection image group sequence and the second detection image group sequence to obtain the detection information set of the middle layer of the grain pile and the detection information set of the bottom layer of the grain pile.

[0049] In some embodiments, the aforementioned execution entity may perform grain pile anomaly detection based on the aforementioned first detection image group sequence and the aforementioned second detection image group sequence to obtain a grain pile middle layer detection information set and a grain pile bottom layer detection information set.

[0050] In some optional implementations of certain embodiments, the execution entity performs grain pile anomaly detection based on the first detection image group sequence and the second detection image group sequence to obtain a grain pile middle layer detection information set and a grain pile bottom layer detection information set, including:

[0051] Step S1: For each first detection image group in the above first detection image group sequence, perform the following steps:

[0052] The first step is to perform image segmentation on each of the first detection images in the first detection image group to obtain a first sub-image sequence set. This sub-image sequence set can be obtained by using a preset target segmentation algorithm.

[0053] As an example, object segmentation algorithms may include, but are not limited to, at least one of the following: Mask Track R-CNN (Region-Cascaded Neural Network) model, Transformer-based video instance segmentation network, and semi-supervised video object segmentation: a video object segmentation neural network model based on Space-Time Correspondence Networks. This ensures that after segmenting the first detection images of consecutive frames, a correspondence still exists between the various first sub-image sequences. This facilitates determining the location region of the same segmented target in the first detection images of different frames.

[0054] The second step is to filter the aforementioned first sub-image sequence set to obtain the first target sub-image group. Among them, for multiple first sub-images in the aforementioned first sub-image sequence set that correspond to the same segmentation target, the first sub-image with the largest image area can be selected as the first target sub-image.

[0055] The third step involves identifying pests in the first target sub-image group to obtain the detection information for the middle layer of the grain pile. This can be achieved using a pre-defined pest identification algorithm. Furthermore, the detection information for the middle layer of the grain pile can include the type and quantity of pests.

[0056] As an example, pest identification algorithms may include, but are not limited to, at least one of the following: decision tree classification algorithm, support vector machine classification algorithm, deep residual network, fast region convolutional neural network, etc.

[0057] Step S2: For each second detection image group in the above second detection image group sequence, perform the following steps:

[0058] The first step is to perform image segmentation on each of the second detection images in the second detection image group to obtain a second sub-image sequence set. This sub-image sequence set can be obtained by performing image segmentation using the target segmentation algorithm described above.

[0059] The second step involves filtering the aforementioned set of second sub-image sequences to obtain a group of second target sub-images. Specifically, for multiple second sub-images in the aforementioned set of second sub-image sequences that correspond to the same segmentation target, the second sub-image with the largest area can be selected as the second target sub-image.

[0060] The third step involves identifying pests in the second target sub-image group to obtain detection information at the bottom of the grain pile. This pest identification algorithm can be used to identify pests in the second target sub-image group to obtain the detection information at the bottom of the grain pile. This detection information may include the type and quantity of pests.

[0061] In practice, considering that pests are often stationary during transport, and that stationary pests are prone to false detection due to residue in the grain pile (e.g., chaff, pest excrement), a method is used. By capturing multiple frames and selecting the first sub-image with the largest area representing each target pest, indicating its active state, this sub-image facilitates pest identification and improves the accuracy of pest identification.

[0062] Step 104: In response to determining that the detection information set of the middle layer of the grain pile and the detection information set of the bottom layer of the grain pile meet the preset abnormal conditions, a three-dimensional statistical map of the grain warehouse is generated.

[0063] In some embodiments, the executing entity may generate a three-dimensional statistical map of the grain warehouse in response to determining that the detection information set of the middle layer of the grain pile and the detection information set of the bottom layer of the grain pile meet preset abnormal conditions. The preset abnormal conditions may include the presence of pest information in the detection information sets of the middle layer of the grain pile and the detection information set of the bottom layer of the grain pile.

[0064] In some optional implementations of certain embodiments, the execution entity generates a three-dimensional statistical map of the grain silo in response to determining that the middle layer detection information set and the bottom layer detection information set of the grain pile meet preset abnormal conditions, including:

[0065] Step S1: Based on the detection depth values ​​corresponding to the aforementioned first detection equipment components, determine the coordinate set of each first detection device in the preset three-dimensional grain silo model. The depth values ​​of each first detection device in the first detection equipment component can be preset based on the grain pile markings and the three-dimensional temperature values ​​of the grain pile. For example, if the grain pile is wheat, the coordinates most suitable for pests to exist on the detectable path below the first detection device can be determined by referring to the three-dimensional temperature values ​​of the grain pile, thus serving as the depth value of the first detection device. This increases the probability of pest capture. Secondly, the three-dimensional grain silo model can be a simulation model pre-established based on the grain silo dimensions and grain pile height, including the position coordinates of each ground vent and each electric telescopic rod. Therefore, the coordinates of the first inspection device corresponding to the detection depth value can be determined in the coordinate system of the three-dimensional grain silo model. Furthermore, the detection depth values ​​of different first detection devices can be different. The maximum difference between the detection depth values ​​of each first detection device does not exceed a preset threshold, meaning that each second detection device only adjusts its detection depth value within the middle layer area.

[0066] Step S2: Based on the location numbers corresponding to the second detection equipment components, determine the coordinate set of each second detection equipment in the three-dimensional grain silo model. Specifically, the location coordinates of the corresponding ventilation opening in the 3D grain silo model can be determined using the corresponding location number of the second detection equipment, and these coordinates are used as the second equipment coordinates.

[0067] Step S3: Based on the first and second equipment coordinate sets, encode the information sets of the middle and bottom layers of the grain pile to obtain the middle layer encoding matrix and the bottom layer encoding matrix. For each layer of detection information, encoding can be performed as follows: First, determine the density value of each pest type within a specified range. This specified range can be a pre-defined range, such as 4 square meters. Then, according to a preset pest type order, determine the corresponding density values ​​as numerical vectors. Finally, insert these numerical vectors into the corresponding first equipment coordinate positions in the matrix as matrix data. This yields the middle layer encoding matrix. Similarly, the bottom layer encoding matrix can be obtained using the second equipment coordinate set and the bottom layer detection information set.

[0068] Step S4: Insert the aforementioned middle-layer encoding matrix and bottom-layer encoding matrix of the grain pile into the aforementioned three-dimensional grain warehouse model to obtain a three-dimensional statistical map of the grain warehouse. Specifically, the data vectors can be inserted into the aforementioned three-dimensional grain warehouse model according to the coordinates corresponding to the data in each matrix of the middle-layer and bottom-layer encoding matrices to obtain the three-dimensional statistical map of the grain warehouse.

[0069] Step 105: Based on the three-dimensional statistical map of the grain warehouse, perform surface path planning for the grain pile to obtain the surface detection path of the grain pile.

[0070] In some embodiments, the aforementioned execution entity can perform surface path planning for the grain pile based on the aforementioned three-dimensional statistical map of the grain warehouse in various ways to obtain the surface detection path of the grain pile.

[0071] In some optional implementations of certain embodiments, the execution entity performs surface path planning on the grain pile based on the aforementioned three-dimensional statistical map of the grain warehouse to obtain a surface detection path for the grain pile, including:

[0072] Step S1 involves performing planar analysis on the aforementioned 3D statistical map of the grain silo to generate a planar pest heat map set. First, the 3D statistical map of the grain silo is divided into upper, middle, and lower layers according to preset boundary lines, resulting in a middle layer dataset and a lower layer dataset. For example, if the grain pile height is 6 meters, then the preset boundary lines for the upper, middle, and lower layers can each occupy a height of 2 meters, i.e., two preset boundary lines of 2 meters and 4 meters. Here, since no pest information was collected for the upper layer, it is not analyzed for the time being. Both the middle and lower layer data include all matrix data of the corresponding level in the 3D statistical map of the grain silo. Missing matrix data can be set to zero. Then, the layer corresponding to the middle layer dataset is further divided according to a preset height value (e.g., 0.5 meters), and the middle layer data in each layer after division is determined as middle layer sub-data, resulting in a middle layer sub-data set for each layer. Next, considering the different vertical coordinates among the various mid-level sub-data sets, for each mid-level sub-data set, the vertical coordinates of the matrix data are removed to map them to planar data, forming a planar data matrix. Then, based on the density values ​​corresponding to each pest type in the planar data matrix, a planar pest heatmap is generated. Here, each pest type in each mid-level sub-data set can correspond to a planar pest heatmap.

[0073] Step S2 involves performing a vertical analysis on the aforementioned three-dimensional statistical map of the grain silo to generate a set of vertical pest heat maps. First, each ventilation trough and its corresponding vent, as well as the first detection device, share the same horizontal coordinate value. Therefore, these can be considered as a single data set. Centered on each horizontal coordinate value, the three-dimensional statistical map of the grain silo is vertically divided into multiple sub-data sets. Then, based on the density values ​​corresponding to each pest type in the sub-data sets, a vertical pest heat map is generated. Similarly, each pest can correspond to one vertical pest heat map, thus obtaining a set of vertical pest heat maps.

[0074] In practice, considering that the first and second detection devices capture pests regionally, data interpolation is unnecessary during data processing; the severity of the pest infestation can be characterized solely by the regional pest density values. This also avoids errors introduced by data interpolation. Then, planar analysis can be used to determine the lateral distribution of pests within the same layer of the grain pile, thereby determining whether the pests are evenly distributed and whether they belong to independent breeding sources. Secondly, vertical analysis can be used to trace the longitudinal correlation of pests along the depth of the grain pile, determining whether pests are spreading across layers, further facilitating the identification of the pest's spread direction and source. Therefore, by combining these two methods, blind spots caused by single-dimensional detection can be avoided, thus improving the accuracy of pest data collection and analysis.

[0075] Step S3: Based on the aforementioned planar and vertical pest heatmap sets, pest areas are located to obtain a pest area information set. This pest area information includes pest area coordinates and pest trend indicators. Here, areas in the planar and vertical pest heatmaps where the pest density exceeds a preset threshold are identified as pest areas. Next, the number of adjacent pest areas is defined as connected components. Here, each connected component represents an area where a type of pest is concentrated. Finally, pest trend indicators are generated based on the size of the pest areas in each planar pest heatmap. For example, if a connected component in an upper-level planar pest heatmap is larger than one in a lower-level planar pest heatmap, an upward trend indicator is generated to represent the upward spread of the pest. Thus, multiple pest trend indicators and corresponding pest areas for each pest type are identified as pest area information.

[0076] Step S4 involves surface path planning for the grain pile based on the pest area information set, resulting in a surface detection path. First, the maximum horizontal bounding circle of the connected components corresponding to pest trend indicators representing upward or downward pest spread is determined. Then, these maximum bounding circles are mapped onto the grain pile surface, resulting in multiple grain pile surface circles. Next, the maximum horizontal bounding rectangle of the connected components corresponding to pest trend indicators representing pest spread in other directions is determined. These maximum bounding rectangles are then mapped onto the grain pile surface, resulting in multiple grain pile surface rectangles. Then, the grain pile surface circles are sorted according to the connected components corresponding to each circle, from largest to smallest, resulting in a grain pile surface circle sequence. Finally, using each grain pile surface circle in the sequence as a waypoint, path planning is performed to obtain the grain pile surface detection path. Here, the constraints for path planning can include ensuring the planned path is within the area containing multiple grain pile surface rectangles and that the planned path is as short as possible. Here, by sorting the circles on the surface of each grain pile, we can prioritize the treatment of areas with more severe insect infestations, thereby quickly determining the extent of the infestation.

[0077] As an example, such as Figure 4 The schematic diagram of the grain pile surface path planning shown illustrates that, in the 3D grain silo model, grain pile surface circles a, b, and c can be obtained. Here, the area of ​​the connected region corresponding to the grain pile surface circles is expressed as: grain pile surface circle a > grain pile surface circle b > grain pile surface circle c. Then, using each grain pile surface circle in the sequence as a path point, path planning is performed on the intelligent grain turning machine 401 to obtain the grain pile surface detection path 402. In addition, the rectangular area on the grain pile surface (not shown) is an optional constraint condition; here, only the shortest planned path is used as the constraint condition.

[0078] Step 106: Based on the grain pile surface detection path and the intelligent turning machine, perform fixed-point sampling of the grain surface to obtain a set of grain surface sampling images.

[0079] In some embodiments, the aforementioned execution entity may perform fixed-point sampling of the grain surface based on the aforementioned grain pile surface detection path and intelligent turning machine to obtain a set of grain surface sampling images.

[0080] In practice, when dealing with pests on the surface of grain, pest traps only passively capture active adult insects. However, due to temperature differences on the grain surface, some preferred pests often lay their eggs there. Therefore, passive pest trapping makes it difficult to detect insect eggs, hindering the early detection of pest distribution and severity. Introducing an intelligent grain turning machine to turn over the grain surface breaks up the compacted layer formed by pests, bringing hidden eggs and adult insects to the surface for photographic identification. This allows for comprehensive identification of abnormalities on the grain surface.

[0081] In some optional implementations of certain embodiments, the aforementioned execution entity performs fixed-point sampling of the grain surface based on the aforementioned grain pile surface detection path and the intelligent grain turning machine to obtain a grain surface sampling image group, including:

[0082] Step S1: Control the intelligent grain pile turning machine to move and turn the grain pile along the aforementioned grain pile surface detection path, and record the turning time points. Specifically, the turning time point when the intelligent grain pile turning machine reaches each grain pile surface circle can be recorded. Furthermore, after reaching each grain pile surface circle, the turning machine can move and turn the grain pile within that area. After completing the turning within the current grain pile surface circle, it moves to the next grain pile surface circle according to the aforementioned grain pile surface detection path.

[0083] Step S2: In response to the time interval between the aforementioned grain turning time point and the current time point being greater than or equal to a preset waiting time, the grain surface camera is controlled to capture images along the aforementioned grain pile surface detection path to obtain a video of the grain pile surface. In practice, considering that pests are prone to entering a "stress state" after the grain turning machine passes by, after a preset waiting time (e.g., 5-10 minutes), the pests can be allowed to transition from a "stress state" (i.e., a static state) to an "exposed state" (i.e., an active state) before being captured, thus maximizing the capture of pests. This greatly reduces the chances of missed or false detections of pests. Here, the grain surface camera can be a high-definition camera equipped with a telephoto lens. Furthermore, the grain surface camera can track and capture images along the historical movement path of the intelligent grain turning machine to obtain a video of the grain pile surface.

[0084] Step S3: Based on the aforementioned pest area information set, perform point-to-point sampling on the surface video of the grain pile to obtain a grain surface sampling image set. Specifically, point-to-point sampling can be performed on the surface video of the grain pile at preset time intervals to obtain the grain surface sampling image set.

[0085] Step 107: Generate grain storage anomaly detection results based on the grain surface sampling image group, the three-dimensional statistical map of the grain storage, and the preset grain storage information.

[0086] In some embodiments, the aforementioned executing entity can generate grain warehouse anomaly detection results based on the aforementioned grain surface sampling image group, the aforementioned three-dimensional statistical map of the grain warehouse, and preset grain warehouse storage information.

[0087] In some optional implementations of certain embodiments, the execution entity generates grain silo anomaly detection results based on the aforementioned grain surface sampling image set, the aforementioned three-dimensional statistical map of the grain silo, and preset grain storage information, including:

[0088] Step S1: Based on the aforementioned grain storage information, pest identification is performed on each grain surface sampling image in the aforementioned grain surface sampling image group to obtain a grain surface pest identification information group. The grain storage information may include a grain type identifier representing the type of stored grain. Next, the aforementioned pest identification algorithm can be used to identify pests in each grain surface sampling image in the aforementioned grain surface sampling image group to obtain a grain surface pest identification information group. Here, the grain surface pest identification information may include the pest type and the corresponding number of pests. Each piece of grain surface pest identification information may correspond to one grain surface sampling image.

[0089] In practice, considering the significant differences in pests affecting different types of stored grains, training the network model simultaneously with all pest species would easily lead to problems due to the large number and similar appearance of pests, hindering training efficiency and improving recognition accuracy. Therefore, for the aforementioned pest identification algorithm, an initial network model with the same structure can be pre-trained based on different stored grain type identifiers and corresponding pest species. The network parameters of the trained model are then stored and a correspondence is established with the stored grain type identifier. Thus, during pest identification, the corresponding network parameters can be selected based on the stored grain type identifier to adjust the pest identification algorithm, resulting in the current algorithm. Pest identification is then performed on each sampled grain surface image in the sampled image group to obtain a set of grain surface pest identification information. This can improve the accuracy of pest detection.

[0090] Step S2 involves inserting the aforementioned grain surface pest identification information group into the aforementioned three-dimensional statistical map of the grain silo to obtain the grain silo anomaly detection result. Specifically, the corresponding image coordinates can be determined based on the grain surface image acquired corresponding to the grain surface pest identification information. This image coordinates can then be transformed into the aforementioned three-dimensional statistical map of the grain silo, and the corresponding grain surface pest identification information can be filled into the corresponding coordinate positions. Finally, the inserted three-dimensional statistical map of the grain silo can be determined as the grain silo anomaly detection result. This improves the accuracy of the grain silo anomaly detection result.

[0091] Optionally, after obtaining the abnormal detection results of the grain pile, the first and second detection equipment components can be reset. This involves emptying the trapping material from both components, refilling them with insect-trapping material, and then repositioning them to the detection location for continued monitoring of pest abnormalities within the grain pile. In practice, resetting the first detection equipment can be achieved by extending the corresponding electric telescopic rod to position it at a designated height within the grain pile. During the extension of the electric telescopic rod, it can also be rotated to counteract the reaction force of the grain pile with its rotational angular momentum, significantly preventing the rod from warping.

[0092] The above-described embodiments of this disclosure have the following beneficial effects: the target anomaly detection method applied to intelligent grain depots through some embodiments of this disclosure can improve the accuracy of grain depot anomaly detection. Specifically, the reason for the large deviation in grain depot anomaly detection results is that different grain types correspond to different pests, and the habits of different pests are not entirely the same, resulting in different distribution depths of pests within the grain pile. Surface sampling methods are difficult to accurately sample pests, making it difficult to determine the actual distribution and severity of pests, thus leading to a large deviation in the anomaly detection results. Based on this, the target anomaly detection method applied to intelligent grain depots through some embodiments of this disclosure firstly retrieves a first detection device component and captures a first detection image sequence corresponding to the first detection device component using a first imaging device. The first detection device component includes at least one first detection device, which is an insect-trapping device laid inside the grain pile. Here, by introducing the first detection device component, it can be used to comprehensively detect pest data inside the grain pile. Simultaneously, the second detection equipment component is retrieved, and a second detection image sequence corresponding to the second detection equipment component is captured by the second imaging device. The second detection equipment component includes at least one second detection device, which is an insect-catching device installed at the ventilation opening of the grain trough. The first detection device corresponds one-to-one with the second detection device. Here, the second detection equipment component can be used to detect pest data at the bottom of the grain pile. Then, based on the aforementioned first and second detection image sequence sequences, grain pile anomaly detection is performed to obtain a middle layer detection information set and a bottom layer detection information set. This can be used to summarize pest information in the middle and bottom layers of the grain pile. Next, in response to determining that the aforementioned middle layer and bottom layer detection information sets meet preset anomaly conditions, a three-dimensional statistical map of the grain warehouse is generated. This can be used to determine the pest distribution within the grain warehouse. Then, based on the aforementioned three-dimensional statistical map of the grain warehouse, a surface path is planned for the grain pile to obtain a surface detection path. Afterwards, based on the aforementioned surface detection path and the intelligent grain turning machine, fixed-point sampling of the grain surface is performed to obtain a grain surface sampling image set. Here, by using surface path planning and fixed-point sampling of the grain pile, not only can supplementary detection be conducted to determine pest information on the grain surface, but it also further improves the monitoring of pest distribution, allowing for more precise control over the severity of pests inside the grain pile. Finally, based on the aforementioned grain surface sampling image set, the aforementioned 3D statistical map of the grain silo, and the preset grain storage information, grain silo anomaly detection results are generated. This significantly improves the accuracy of grain silo anomaly detection results.

[0093] Further reference Figure 5 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a target anomaly detection device applied to intelligent grain depots. These device embodiments are similar to... Figure 1Corresponding to the method embodiments shown, the target anomaly detection device applied to smart grain depots can be specifically applied to various electronic devices.

[0094] like Figure 5 As shown, a target anomaly detection device 500 applied to a smart grain depot in some embodiments includes: a first retrieval and imaging unit 501, a second retrieval and imaging unit 502, a grain pile anomaly detection unit 503, a first generation unit 504, a detection path planning unit 505, a grain surface fixed-point sampling unit 506, and a second generation unit 507. The first retrieval and imaging unit 501 is configured to retrieve a first detection device component and capture a first detection image sequence corresponding to the first detection device component using a first imaging device. The first detection device component includes at least one first detection device, which is an insect trapping device laid inside the grain pile. The second retrieval and imaging unit 502 is configured to retrieve a second detection device component and capture a second detection image sequence corresponding to the second detection device component using a second imaging device. The second detection device component includes at least one second detection device, which is an insect trapping device installed at the ventilation opening of the trough. The first detection device and the second detection device correspond one-to-one. The grain pile anomaly detection unit 503 is configured to detect the first detection image sequence... The grain pile anomaly detection is performed using the above-mentioned second detection image group sequence to obtain a grain pile middle layer detection information set and a grain pile bottom layer detection information set; the first generation unit 504 is configured to generate a grain warehouse three-dimensional statistical map in response to determining that the above-mentioned grain pile middle layer detection information set and the above-mentioned grain pile bottom layer detection information set meet preset anomaly conditions; the detection path planning unit 505 is configured to perform grain pile surface path planning based on the above-mentioned grain warehouse three-dimensional statistical map to obtain a grain pile surface detection path; the grain surface fixed-point sampling unit 506 is configured to perform grain surface fixed-point sampling based on the above-mentioned grain pile surface detection path and the intelligent grain turning machine to obtain a grain surface sampling image group; the second generation unit 507 is configured to generate a grain warehouse anomaly detection result based on the above-mentioned grain surface sampling image group, the above-mentioned grain warehouse three-dimensional statistical map and preset grain warehouse storage information.

[0095] It is understandable that the units described in the target anomaly detection device 500 applied to intelligent grain depots are related to the reference... Figure 1 The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method are also applicable to the target anomaly detection device 500 and its constituent units applied to intelligent grain depots, and will not be repeated here.

[0096] The following is for reference. Figure 6 It illustrates a schematic diagram of the structure of an electronic device (such as a computing device) suitable for implementing some embodiments of the present disclosure. Figure 6The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments of this disclosure. Figure 6 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory may include a non-volatile storage medium and internal memory. The non-volatile storage medium may store an operating system and a computer program. The computer program includes program instructions that, when executed, cause the processor to perform any of the methods described above. The processor provides computational and control capabilities to support the operation of the entire computer device. The internal memory provides an environment for the execution of the computer program in the non-volatile storage medium; when executed by the processor, the computer program causes the processor to perform any of the methods described above. The network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the computer device to which the present disclosure is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0097] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0098] In one embodiment, the processor is configured to run a computer program stored in a memory to perform the following steps: retrieving a first detection device component and capturing a first detection image sequence corresponding to the first detection device component using a first imaging device, wherein the first detection device component includes at least one first detection device, which is an insect trapping device installed inside the grain pile; retrieving a second detection device component and capturing a second detection image sequence corresponding to the second detection device component using a second imaging device, wherein the second detection device component includes at least one second detection device, which is an insect trapping device installed at the ventilation opening of the trough, and the first detection device and... The second detection device is paired with the first detection image group sequence and the second detection image group sequence to perform grain pile anomaly detection, obtaining a grain pile middle layer detection information set and a grain pile bottom layer detection information set; in response to determining that the grain pile middle layer detection information set and the grain pile bottom layer detection information set meet the preset anomaly conditions, a three-dimensional statistical map of the grain warehouse is generated; based on the grain warehouse three-dimensional statistical map, a grain pile surface path is planned to obtain a grain pile surface detection path; based on the grain pile surface detection path and the intelligent grain turning machine, fixed-point sampling of the grain surface is performed to obtain a grain surface sampling image group; based on the grain surface sampling image group, the grain warehouse three-dimensional statistical map, and the preset grain warehouse storage information, a grain warehouse anomaly detection result is generated.

[0099] This disclosure also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, and the method implemented when the program instructions are executed can be referred to the various embodiments of the methods described above.

[0100] The aforementioned computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. Alternatively, the aforementioned computer-readable storage medium may be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0101] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0102] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A target anomaly detection method applied to intelligent grain depots, characterized in that, include: The first detection equipment component is recovered, and a first detection image sequence corresponding to the first detection equipment component is captured by a first imaging device. The first detection equipment component includes at least one first detection device, which is an insect trapping device laid inside the grain pile. The second detection device component is recovered, and a second detection image sequence corresponding to the second detection device component is captured by the second imaging device. The second detection device component includes at least one second detection device, which is an insect trapping device installed at the ventilation opening of the trench. The first detection device and the second detection device correspond one-to-one. Based on the first and second detection image group sequences, grain pile anomaly detection is performed to obtain the middle layer detection information set and the bottom layer detection information set of the grain pile. In response to determining that the detection information set of the middle layer of the grain pile and the detection information set of the bottom layer of the grain pile meet the preset abnormal conditions, a three-dimensional statistical map of the grain warehouse is generated. Based on the three-dimensional statistical map of the grain warehouse, a path planning is performed on the surface of the grain pile to obtain the detection path on the surface of the grain pile. Based on the grain pile surface detection path and the intelligent turning machine, fixed-point sampling of the grain surface is performed to obtain a group of grain surface sampling images. Based on the sampled grain surface image group, the three-dimensional statistical map of the grain warehouse, and the preset grain storage information, anomaly detection results for the grain warehouse are generated.

2. The method according to claim 1, characterized in that, The first detection devices in the first detection equipment assembly are all fixedly connected to the first end of the electric telescopic pole, and the second end of the electric telescopic pole is fixed to the top of the grain silo. The process of retrieving the first detection equipment assembly and capturing a first sequence of detection images corresponding to the first detection equipment assembly using the first imaging device includes: For each first detection device in the first detection device assembly, perform the following steps: Control the electric telescopic pole corresponding to the first detection device to retrieve the first detection device, and control the automatic guide vehicle to retrieve the captured items in the first detection device and return to the first shooting area, wherein the automatic guide vehicle is equipped with a cargo pallet; The first imaging device captures images of the first imaging area to obtain a first detection image group.

3. The method according to claim 2, characterized in that, The second detection device in the second detection equipment assembly is fixed on a tracked remote-controlled transport vehicle. The track of the remote-controlled transport vehicle is laid in the ventilation trough of the grain silo. Each second detection device in the second detection equipment assembly has a corresponding location number. The process of recovering the second detection equipment assembly and capturing a sequence of second detection images corresponding to the second detection equipment assembly using a second imaging device includes: According to the location number corresponding to the second testing device, perform the following steps for each second testing device included in the second testing device component: Control the corresponding remote-controlled transport vehicle to retrieve the second detection device, and place the captured device from the second detection device into the second shooting area; The second imaging device captures images of the second imaging area to obtain a second detection image group.

4. The method according to claim 3, characterized in that, The step of performing anomaly detection on the grain pile based on the first detection image group sequence and the second detection image group sequence to obtain a middle layer detection information set and a bottom layer detection information set of the grain pile includes: For each first detection image group in the first detection image group sequence, perform the following steps: Image segmentation is performed on each of the first detection images in the first detection image group to obtain a first sub-image sequence set; The first sub-image sequence set is filtered to obtain the first target sub-image group; Pest identification is performed on the first target sub-image group to obtain the detection information of the middle layer of the grain pile; For each second detection image group in the second detection image group sequence, perform the following steps: Image segmentation is performed on each second detection image in the second detection image group to obtain a second sub-image sequence set; The second sub-image sequence set is filtered to obtain the second target sub-image group; Pest identification is performed on the second target sub-image group to obtain the detection information of the bottom layer of the grain pile.

5. The method according to claim 4, characterized in that, The step of generating a three-dimensional statistical map of the grain warehouse in response to determining that the detection information set of the middle layer of the grain pile and the detection information set of the bottom layer of the grain pile meet preset abnormal conditions includes: Based on the detection depth value corresponding to the first detection equipment component, determine the first equipment coordinate set corresponding to each first detection equipment in the preset three-dimensional grain warehouse model; Based on the location number corresponding to the second detection equipment component, determine the second equipment coordinate set corresponding to each second detection equipment in the three-dimensional grain warehouse model; Based on the first device coordinate set and the second device coordinate set, information encoding is performed on the middle layer detection information set and the bottom layer detection information set of the grain pile to obtain the middle layer encoding matrix and the bottom layer encoding matrix of the grain pile. The middle layer encoding matrix and the bottom layer encoding matrix of the grain pile are inserted into the three-dimensional grain warehouse model to obtain a three-dimensional statistical map of the grain warehouse.

6. The method according to claim 5, characterized in that, The step of planning the surface path of the grain pile based on the three-dimensional statistical map of the grain warehouse to obtain the surface detection path of the grain pile includes: A planar analysis was performed on the three-dimensional statistical map of the grain warehouse to generate a set of planar insect pest heat maps; A vertical analysis was performed on the three-dimensional statistical map of the grain warehouse to generate a set of vertical pest heat maps; The pest area is located based on the planar pest heat map set and the vertical pest heat map set to obtain a pest area information set, wherein the pest area information includes: pest area coordinates and pest trend identifier. Based on the pest area information set, a surface path planning is performed on the grain pile to obtain the surface detection path of the grain pile.

7. The method according to claim 6, characterized in that, The process of performing fixed-point sampling of the grain surface based on the grain pile surface detection path and the intelligent turning machine to obtain a set of grain surface sampling images includes: Control the intelligent grain turning machine to move and turn the grain along the detection path on the surface of the grain pile, and record the turning time points; In response to a time interval between the turning point and the current time being greater than or equal to a preset waiting time, the grain surface camera is controlled to capture images along the detection path on the surface of the grain pile to obtain a video of the grain pile surface. Based on the pest area information set, the surface video of the grain pile is sampled at fixed points to obtain a set of sampled grain surface images.

8. The method according to claim 7, characterized in that, The step of generating grain warehouse anomaly detection results based on the grain surface sampling image group, the three-dimensional statistical map of the grain warehouse, and preset grain warehouse storage information includes: Based on the grain storage information, pest identification is performed on each grain surface sampling image in the grain surface sampling image group to obtain a grain surface pest identification information group. The grain surface pest identification information group is inserted into the three-dimensional statistical map of the grain warehouse to obtain the grain warehouse anomaly detection results.

9. A target anomaly detection device applied to intelligent grain depots, characterized in that, include: The first recovery and imaging unit is configured to recover the first detection device component and to capture a first detection image sequence corresponding to the first detection device component through the first imaging device, wherein the first detection device component includes at least one first detection device, which is an insect trapping device laid inside the grain pile. The second recovery and imaging unit is configured to recover the second detection device component and to capture a second detection image sequence corresponding to the second detection device component through the second imaging device. The second detection device component includes at least one second detection device, which is an insect trapping device installed at the ventilation opening of the trench. The first detection device and the second detection device correspond one-to-one. The grain pile anomaly detection unit is configured to perform grain pile anomaly detection based on the first detection image group sequence and the second detection image group sequence to obtain a grain pile middle layer detection information set and a grain pile bottom layer detection information set. The first generation unit is configured to generate a three-dimensional statistical map of the grain warehouse in response to determining that the middle layer detection information set of the grain pile and the bottom layer detection information set of the grain pile meet preset abnormal conditions. The detection path planning unit is configured to perform surface path planning for the grain pile based on the three-dimensional statistical map of the grain warehouse, and obtain the surface detection path of the grain pile. The grain surface fixed-point sampling unit is configured to perform fixed-point sampling of the grain surface based on the detection path of the grain pile surface and the intelligent turning machine, and obtain a grain surface sampling image group; The second generation unit is configured to generate grain warehouse anomaly detection results based on the grain surface sampling image group, the grain warehouse three-dimensional statistical map, and preset grain warehouse storage information.

10. An electronic device, characterized in that, include: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-8.

Images