Rare metal ore sorting method, apparatus, device, and storage medium

Abstract

The application discloses a rare metal ore sorting method, device, equipment and storage medium, and relates to the technical field of ore sorting. The rare metal ore sorting method comprises the following steps: acquiring an ore picture, and sorting an ore corresponding to the ore picture by using a preset ore sorting model. The ore sorting model is trained based on rare metal ore pictures with different lithofacies, so as to sort the rare metal ore based on the lithofacies characteristics of the ore. For a rare metal ore with complex lithofacies characteristics, the ore sorting model is trained based on rare metal ore pictures with different lithofacies. By using the trained ore sorting model, the sorting system can be adjusted according to the lithofacies characteristics of the ore when sorting the ore, the detection accuracy of the rare metal ore is improved, and the metal recovery rate of the rare metal ore sorting is improved.

Description

Technical Field

[0001] This application relates to the field of ore sorting technology, and in particular to methods, apparatus, equipment and storage media for sorting rare metal ores. Background Technology

[0002] Ore sorting is the process of separating symbiotic ores, and it can effectively improve the utilization efficiency of mineral resources. Current rare metal ore sorting methods usually rely on the physicochemical properties of the ore for sorting. However, due to the complex petrographic characteristics of rare metal ores, current methods result in low metal recovery rates. Summary of the Invention

[0003] The main objective of this application is to provide a method, apparatus, equipment, and storage medium for sorting rare metal ores, aiming to solve the technical problem of low metal recovery rate in the sorting of rare metal ores.

[0004] To achieve the above objectives, this application proposes a method for separating rare metal ores, the method comprising:

[0005] Obtain ore images;

[0006] The ore corresponding to the ore image is sorted by a preset ore sorting model. The ore sorting model is trained based on rare metal ore sample images with different lithofacies characteristics to sort rare metal ores based on the lithofacies characteristics of the ore.

[0007] In one embodiment, prior to the step of acquiring the ore image, the method further includes:

[0008] Obtain a first ore dataset, wherein the first ore dataset consists of ore sample images corresponding to ores with radii smaller than a preset threshold;

[0009] The ore sample images in the first ore dataset are detected based on the pre-trained model to obtain a first detection result, wherein the first detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore.

[0010] Based on the first detection result, the preset parameters to be updated in the pre-trained model are adjusted to obtain the ore sorting model, wherein the parameters other than the parameters to be updated are frozen during the training process of the ore sorting model.

[0011] In one embodiment, the ore sorting model has a channel attention module and a spatial attention module, and the step of sorting the ore corresponding to the ore image using the preset ore sorting model includes:

[0012] The channel attention module and the spatial attention module are used to extract features from the ore image to obtain the lithofacies features;

[0013] Based on the lithofacies characteristics and the detection layer of the ore sorting model, a target detection result is obtained, wherein the target detection result includes whether the corresponding ore is a rare metal ore or not.

[0014] Based on the target detection results, rare metal ores are sorted.

[0015] In one embodiment, the step of extracting features from the ore image using the channel attention module and the spatial attention module to obtain the lithofacies features includes:

[0016] Based on the channel attention module, channel features are extracted from the ore image to obtain a channel feature map;

[0017] Based on the spatial attention module, max pooling and average pooling are performed on the channel feature map to obtain max pooling features and average pooling features.

[0018] The max pooling feature and the average pooling feature are concatenated to obtain the concatenation feature;

[0019] Perform a convolution operation on the connection features to obtain the feature map spatial weights;

[0020] The lithofacies features are obtained by dot product of the spatial weights of the feature map and the channel feature map.

[0021] In one embodiment, after the step of acquiring the ore image, the method further includes:

[0022] The ore image is preprocessed to obtain a preprocessed image;

[0023] The preprocessed images of a preset number are stitched together to obtain the stitched image.

[0024] The pre-set ore sorting model is used to sort the ore corresponding to the stitched image.

[0025] In one embodiment, prior to the step of acquiring the ore image, the method further includes:

[0026] Obtain a second ore dataset, wherein the second ore dataset consists of ore images corresponding to ores of different grades, and among the ores of different grades, the number of low-grade ores is higher than a preset quantity threshold, and the low-grade ores are ores with a target metal content lower than a preset content threshold;

[0027] The pre-trained model is used to detect the ore sample images in the second ore dataset to obtain a second detection result, wherein the second detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore.

[0028] Based on the second detection result, the parameters in the pre-trained model are adjusted to obtain the ore sorting model.

[0029] In one embodiment, before the step of acquiring the ore image, the method further includes:

[0030] Obtain a third ore dataset, wherein the third ore dataset consists of ore sample images corresponding to the ore in the target mining area, and the ore in the target mining area consists of associated minerals and target metals in the target mining area;

[0031] The ore sample images in the third ore dataset are detected based on the pre-trained model to obtain a third detection result, wherein the third detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore.

[0032] Based on the third detection result, the parameters in the pre-trained model are adjusted to obtain the ore sorting model.

[0033] Furthermore, to achieve the above objectives, this application also proposes a rare metal ore sorting device, which includes:

[0034] The image acquisition module is used to acquire images of minerals;

[0035] The ore sorting module is used to sort the ore corresponding to the ore image using a preset ore sorting model. The ore sorting model is trained based on rare metal ore sample images with different lithofacies characteristics to sort rare metal ores based on the lithofacies characteristics of the ore.

[0036] In addition, to achieve the above objectives, this application also proposes a rare metal ore sorting device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the rare metal ore sorting method described above.

[0037] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the rare metal ore sorting method described above.

[0038] One or more technical solutions proposed in this application have at least the following technical effects:

[0039] Obtain ore images and sort the ore corresponding to the ore images using a preset ore sorting model. The ore sorting model is trained based on rare metal ore images with different lithofacies to sort rare metal ores based on the lithofacies characteristics of the ore.

[0040] For rare metal ores with complex lithofacies characteristics, this application trains an ore sorting model based on images of rare metal ores with different lithofacies. The ore sorting model obtained through training enables the sorting system to be adjusted according to the lithofacies characteristics of the ore during ore sorting, thereby improving the detection accuracy of rare metal ores and thus improving the metal recovery rate of rare metal ore sorting. Attached Figure Description

[0041] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 A schematic flowchart of an embodiment of the rare metal ore sorting method of this application;

[0044] Figure 2 This is a schematic flowchart of Embodiment 2 of the rare metal ore sorting method of this application;

[0045] Figure 3 This is a schematic flowchart of Embodiment 3 of the rare metal ore sorting method of this application;

[0046] Figure 4 This is a schematic diagram of the module structure of the rare metal ore sorting device according to an embodiment of this application;

[0047] Figure 5 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the rare metal ore sorting method in the embodiments of this application.

[0048] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0049] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0050] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0051] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, a rare metal ore sorting device, etc. The following description uses a rare metal ore sorting device as an example to illustrate this embodiment and the subsequent embodiments.

[0052] Based on this, the embodiments of this application provide a method for separating rare metal ores, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the rare metal ore sorting method of this application.

[0053] In this embodiment, the rare metal ore sorting method includes steps S10 to S20:

[0054] Step S10: Obtain an image of the ore;

[0055] It should be noted that the ore images are those of the ore after crushing and screening.

[0056] Understandably, most ores consist of valuable rare metals and useless gangue. Therefore, it is necessary to crush and screen the ores to separate the valuable rare metals from the gangue as much as possible, thus providing favorable conditions for subsequent rare metal sorting.

[0057] In one feasible implementation, the specific implementation after obtaining the ore image may also be:

[0058] The ore image is preprocessed to obtain a preprocessed image. Multiple preprocessed images of a preset number are then stitched together to obtain a stitched image. The ore corresponding to the stitched image is then sorted using a preset ore sorting model.

[0059] It should be noted that image preprocessing operations include distortion correction, brightness adjustment (overexposure, underexposure), white balance, etc. The preset stitching quantity can be set according to specific needs. The preset stitching quantity is usually 2, 4, or 8. The image stitching operation is to stitch multiple images together into a larger image.

[0060] It is understandable that the shooting conditions of mineral images will affect the quality of the resulting images. For example, mineral images taken under different lighting conditions may exhibit different color representations. In such cases, white balance adjustments can be used to maintain color consistency across different images. Therefore, this embodiment requires preprocessing the captured mineral images to make the important features in the images more prominent and consistent, thereby enabling the detection and classification of rare metal ores using the preprocessed images.

[0061] Due to the massive parallel processing capabilities of high-performance computing hardware such as GPUs, using only a single image for detection cannot fully utilize these capabilities. Therefore, this embodiment stitches together multiple images and uses the larger, stitched image for detection. This avoids the hardware idleness caused by processing small images, thereby improving the utilization of computing resources and increasing detection efficiency.

[0062] In one embodiment, a single high-performance image recognition server can be configured to simultaneously provide image processing and analysis services to multiple production lines. This avoids the need for each production line to have its own independent, high-cost computing device, thereby achieving effective sharing of hardware resources, reducing direct equipment costs, and lowering subsequent maintenance costs. Furthermore, the centralized image recognition server facilitates unified management and scheduling, allowing for dynamic allocation of computing resources based on the actual needs of each production line, thus improving the utilization rate of hardware resources.

[0063] Step S20: The ore corresponding to the ore image is sorted using a preset ore sorting model. The ore sorting model is trained based on rare metal ore sample images with different lithofacies characteristics to sort rare metal ores based on the lithofacies characteristics of the ore.

[0064] It should be noted that the petrographic characteristics of an ore are the combination, structure, and distribution of mineral components within the ore.

[0065] Understandably, rare metal ores have complex petrographic characteristics. Existing sorting equipment usually relies on the physicochemical properties (density and magnetism) of the ore for sorting. However, these properties fail to fully consider the petrographic combination of the ore, making traditional sorting methods (gravity separation and flotation) prone to low metal recovery rates.

[0066] To improve metal recovery rates in ore sorting, the petrographic characteristics of the ore must be fully considered during crushing, screening, and sorting operations. Therefore, this embodiment trains an ore sorting model using images of rare metal ores with different petrographic characteristics, allowing the model to learn about different petrographic phases of rare metal ores during training. Thus, the trained ore sorting model can adaptively adjust based on the petrographic characteristics of the ore during sorting, improving the detection accuracy of rare metal ores and thereby increasing the metal recovery rate.

[0067] In one feasible implementation, the ore sorting model has a channel attention module and a spatial attention module. The specific implementation of sorting the ore corresponding to the ore image using the preset ore sorting model can also be:

[0068] The channel attention module and the spatial attention module are used to extract features from the ore image to obtain the lithofacies features. Based on the lithofacies features and the detection layer of the ore sorting model, detection is performed to obtain the target detection result. The target detection result includes whether the corresponding ore is a rare metal ore or not. Based on the target detection result, the rare metal ore is sorted.

[0069] It should be noted that the channel attention module and the spatial attention module respectively obtain the importance of each feature channel and each feature space in the input feature layer, and use the obtained importance to enhance important features and suppress features that are not important to the current task.

[0070] Understandably, some rare metal ores have complex lithofacies characteristics, with various minerals mixed together in the form of fine particles. In order to accurately separate the target rare metal from ores with complex lithofacies characteristics, it is necessary to pay attention to the important characteristics of the ore.

[0071] When processing ore images, the channel attention module can learn which channels are more important and automatically weight feature channels that are more helpful for classification, while the spatial attention module focuses on highlighting the most important spatial locations or regions in the image. Therefore, this embodiment, through the channel attention module and the spatial attention module, can focus on the key features of key regions in rare metal ores, improve the accuracy of rare metal ore sorting, thereby reducing the under-sorting of rare metal ores and reducing ore loss.

[0072] In one feasible implementation, the specific implementation of extracting features from the ore image through the channel attention module and the spatial attention module to obtain the lithofacies features can also be:

[0073] Based on the channel attention module, channel features are extracted from the ore image to obtain a channel feature map. Based on the spatial attention module, max pooling and average pooling are performed on the channel feature map to obtain max pooling features and average pooling features. The max pooling features and average pooling features are concatenated to obtain connection features. Convolution operation is performed on the connection features to obtain feature map spatial weights. The feature map spatial weights and the channel feature map are dot-producted to obtain lithofacies features.

[0074] It is understandable that different mineral components in an ore will produce different features in different channels of an ore image. Therefore, this embodiment uses a channel attention module to make the model focus on important channels in the ore image and connects the channel features after average pooling and max pooling in order to highlight the most significant channel feature points while preserving global structural information.

[0075] Spatial convolution of connectivity features allows the model to focus not only on important channels in ore images but also on important regions. By performing dot product operations between the calculated feature map spatial weights and the original channel feature maps, the final lithofacies features contain key information from both the channel layer and important lithofacies regions, thereby improving the rare metal ore detection capability of the trained model.

[0076] In summary, this embodiment acquires ore images and sorts the ore corresponding to the ore images using a preset ore sorting model. The ore sorting model is trained based on rare metal ore images with different lithofacies, and sorts rare metal ores based on the lithofacies characteristics of the ore.

[0077] For rare metal ores with complex lithofacies characteristics, this embodiment trains an ore sorting model based on images of rare metal ores with different lithofacies. The ore sorting model obtained through training enables the sorting system to be adjusted according to the lithofacies characteristics of the ore during ore sorting, thereby improving the detection accuracy of rare metal ores and thus improving the metal recovery rate of rare metal ore sorting.

[0078] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Before step S10, the rare metal ore sorting method further includes steps S01 to S03:

[0079] Step S01: Obtain the first ore dataset, wherein the first ore dataset consists of ore sample images corresponding to ores with radii smaller than a preset threshold;

[0080] It should be noted that the preset threshold is the radius threshold of the rare metal ore after crushing, and can be set according to specific needs. In this embodiment, the preset threshold is set to 20mm.

[0081] It is understandable that the radius of rare metal ores after crushing is between 20mm and 150mm, with some small-particle ores having a radius of less than 20mm. When using object detection methods to detect these small-particle ores, the model's accuracy may be low due to insufficient image detail capture capabilities. Therefore, this embodiment trains the model using a small-particle ore dataset to improve its detection capability.

[0082] Step S02: Detect the ore sample images in the first ore dataset based on the pre-trained model to obtain a first detection result, wherein the first detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore.

[0083] It should be noted that the pre-trained model is a model obtained after training on a large object detection dataset.

[0084] It is understandable that the target detection dataset used for pre-training contains a small number of small target objects. Therefore, in this embodiment, a small-granularity rare mineral dataset is used to further train the pre-trained model for the task of small target mineral detection, so as to obtain an ore sorting model.

[0085] Since pre-trained models are trained on large object detection datasets, they can recognize various common features such as edges and textures. Therefore, training based on pre-trained models can reduce the resource consumption required for training and improve the accuracy of the final model.

[0086] Step S03: Based on the first detection result, adjust the preset parameters to be updated in the pre-trained model to obtain the ore sorting model, wherein the parameters other than the parameters to be updated are frozen during the training process of the ore sorting model.

[0087] It should be noted that the frozen layers are those containing the general features of the model and those that are not sensitive to small target datasets. In this embodiment, the shallow layers of the neural network near the input end are frozen, as well as the deep layers of the neural network that are not sensitive to small target datasets.

[0088] Understandably, in pre-trained deep learning models, the shallow layers typically learn more basic and general features. By freezing these layers, it can be ensured that the model does not lose the learned general features, thus maintaining its performance on the task of detecting small-grained rare metal ores.

[0089] Furthermore, since this embodiment fine-tunes the pre-trained model using a small-granular dataset, comprehensive fine-tuning of the entire network could lead to model instability. Therefore, this embodiment avoids unnecessary parameter updates and maintains model stability by freezing shallow layers of the neural network and layers insensitive to small target datasets during training. Moreover, because only a small number of parameters participate in training, the number of parameters requiring optimization is reduced, allowing the model to reach convergence more quickly and improving training efficiency.

[0090] In summary, this embodiment obtains a first ore dataset, which consists of ore sample images corresponding to ores with radii smaller than a preset threshold. Based on a pre-trained model, the ore sample images in the first ore dataset are detected to obtain a first detection result. This first detection result includes whether the corresponding ore is a rare metal ore or not. Based on the first detection result, preset parameters to be updated in the pre-trained model are adjusted to obtain an ore sorting model. Parameters other than those to be updated are frozen during the training process of the ore sorting model.

[0091] The pre-trained model is obtained by training on a large object detection dataset. Therefore, this embodiment improves the model's detection accuracy for small-grained ores by fine-tuning the model using a small-grained rare mineral dataset based on the pre-trained model. Furthermore, since this embodiment fine-tunes the model using a small-grained dataset based on the pre-trained model, to avoid model instability caused by comprehensive fine-tuning, this embodiment freezes shallow layers containing general features and layers insensitive to small object datasets during fine-tuning. This improves the stability of the model during fine-tuning and increases the model's training efficiency.

[0092] Based on the first and second embodiments of this application, in the third embodiment of this application, the content that is the same as or similar to that in embodiments one and two above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 Before step S10, the rare metal ore sorting method further includes steps A01 to A03:

[0093] Step A01: Obtain the second ore dataset, wherein the second ore dataset is composed of ore images corresponding to different grades of ore, and the number of low-grade ore in the different grades of ore is higher than a preset quantity threshold, wherein the low-grade ore is ore with a target metal content lower than a preset content threshold;

[0094] It should be noted that the preset quantity threshold and the preset content threshold are set based on specific task requirements.

[0095] Understandably, actual ore sorting processes require separating ores of different grades, and low-grade and high-grade ores typically have different characteristics. A model trained on high-grade ores cannot accurately detect low-grade ores. Therefore, to enable the trained ore sorting model to generalize to sort ores of different grades, it needs to be trained using images of ores of varying grades.

[0096] Furthermore, since low-grade ore is typically more abundant than high-grade ore, the quantity of low-grade ore needs to exceed a preset threshold to ensure the trained model can improve the detection and sorting accuracy of low-grade ore. This embodiment also prevents overfitting by training the model using images of ore of different grades.

[0097] Step A02: Detect the ore sample images in the second ore dataset based on the pre-trained model to obtain a second detection result, wherein the second detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore.

[0098] Understandably, the pre-trained model is a model that has already learned the general features of the object detection task. In this embodiment, based on the pre-trained model, fine-tuning the pre-training is performed using rare metal ores of different grades, and the number of low-grade ores is made higher than a preset threshold. This allows the model to learn the features of ores of different grades, thereby improving the detection and sorting accuracy of low-grade ores and preventing overfitting.

[0099] Step A03: Based on the second detection result, adjust the parameters in the pre-trained model to obtain the ore sorting model.

[0100] In one feasible implementation, the specific implementation method prior to acquiring the ore image may also be:

[0101] A third ore dataset is obtained, which consists of ore sample images corresponding to the ore in the target mining area. The ore in the target mining area consists of associated minerals and target metals. The ore sample images in the third ore dataset are detected based on a pre-trained model to obtain a third detection result. The third detection result includes whether the corresponding ore is a rare metal ore or not. Based on the third detection result, the parameters in the pre-trained model are adjusted to obtain an ore sorting model.

[0102] It is understood that this embodiment is based on the petrographic characteristics of the ore to detect and sort rare metal ores. Geological conditions vary in different regions, therefore, the types of associated minerals and the appearance characteristics of rare metal ores differ in different regions. That is, the petrographic characteristics of the ore are related to the region of the ore and vein.

[0103] Therefore, in this embodiment, when sorting rare metal ores, the model is specifically trained using ore sample images from the target mining area, enabling the model to better learn the lithofacies characteristics of the ore in the target mining area. The ore sorting model obtained through this training can improve the accuracy of rare metal ore detection when used to detect ore in the target mining area.

[0104] In one feasible implementation, the specific implementation method prior to acquiring the ore image may also be:

[0105] A background dataset and an ore sample dataset are obtained. The background dataset consists of images of conveyor belts transporting ore. Based on a pre-trained model, the images in the background dataset and the ore sample dataset are detected to obtain a fourth detection result. The fourth detection result includes whether the corresponding ore is a rare metal ore or not. Based on the fourth detection result, the parameters in the pre-trained model are adjusted to obtain the ore sorting model.

[0106] It is understandable that the background of rare metal ore sorting is relatively simple, making it impossible for the ore sorting model to obtain relevant information about the ore from the background. Therefore, more reasoning is required when sorting the ore, which increases the computing power of the image recognition server.

[0107] Therefore, this embodiment increases the background dataset and performs fine-tuning pre-training using the background dataset, enabling the trained ore sorting model to detect ore using the background information of ore images. This reduces the computational load of image detection and image classification algorithm inference, and correspondingly reduces the computing power consumption of the image recognition server.

[0108] In summary, this embodiment obtains a second ore dataset, which consists of ore images corresponding to ores of different grades. Among the ores of different grades, the number of low-grade ores is higher than a preset quantity threshold. The low-grade ores are ores with a target metal content lower than a preset content threshold. Based on a pre-trained model, the ore sample images in the second ore dataset are detected to obtain a second detection result. The second detection result includes whether the corresponding ore is a rare metal ore or not. Based on the second detection result, the parameters in the pre-trained model are adjusted to obtain an ore sorting model.

[0109] In actual ore sorting processes, the rare metal ores to be sorted are from different mining areas and of different grades, each with corresponding petrographic characteristics. Therefore, this embodiment trains the model using ore datasets from different mining areas and of different grades, enabling the model to better learn the petrographic characteristics of the corresponding ores, thereby improving the detection accuracy of the ore sorting model for rare metal ores. Furthermore, this embodiment also fine-tunes the model by adding a background dataset, allowing the model to sort ore using background information, thus reducing the computational power consumption of the image recognition server.

[0110] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the rare metal ore sorting method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0111] This application also provides a rare metal ore sorting device; please refer to... Figure 4 The rare metal ore sorting device includes:

[0112] Image acquisition module 10 is used to acquire images of ore.

[0113] The ore sorting module 20 is used to sort the ore corresponding to the ore image using a preset ore sorting model. The ore sorting model is trained based on rare metal ore sample images with different lithofacies characteristics to sort rare metal ores based on the lithofacies characteristics of the ore.

[0114] In one embodiment, the rare metal ore sorting device further includes:

[0115] The first dataset acquisition module is used to acquire a first ore dataset, wherein the first ore dataset consists of ore sample images corresponding to ores with radii smaller than a preset threshold;

[0116] The first detection module is used to detect ore sample images in the first ore dataset based on a pre-trained model to obtain a first detection result, wherein the first detection result includes whether the corresponding ore is a rare metal ore or whether the corresponding ore is not a rare metal ore.

[0117] The first parameter adjustment module is used to adjust the preset parameters to be updated in the pre-trained model based on the first detection result to obtain the ore sorting model, wherein the parameters other than the parameters to be updated are frozen during the training process of the ore sorting model.

[0118] In one embodiment, the ore sorting module further includes:

[0119] The feature extraction submodule is used to extract features from the ore image through the channel attention module and the spatial attention module to obtain the lithofacies features;

[0120] The ore detection submodule is used to perform detection based on the lithofacies characteristics and the detection layer of the ore sorting model to obtain the target detection result, wherein the target detection result includes whether the corresponding ore is a rare metal ore or whether the corresponding ore is not a rare metal ore.

[0121] The ore sorting submodule is used to sort rare metal ores based on the target detection results.

[0122] In one embodiment, the feature extraction submodule further includes:

[0123] The channel feature extraction unit is used to extract channel features from the ore image based on the channel attention module to obtain a channel feature map;

[0124] The spatial feature extraction unit is used to perform max pooling and average pooling on the channel feature map based on the spatial attention module to obtain max pooling features and average pooling features.

[0125] A feature connection unit is used to connect the max pooling feature and the average pooling feature to obtain a connection feature;

[0126] The weight calculation unit is used to perform convolution operations on the connection features to obtain the feature map space weights;

[0127] The feature dot product unit is used to perform a dot product between the spatial weights of the feature map and the channel feature map to obtain the lithofacies features.

[0128] In one embodiment, the rare metal ore sorting device further includes:

[0129] The image processing module is used to preprocess the ore image to obtain a preprocessed image;

[0130] The image stitching module is used to stitch together multiple preprocessed images of a preset number to obtain a stitched image;

[0131] The splicing and sorting module sorts the ores corresponding to the spliced ​​images using a preset ore sorting model.

[0132] In one embodiment, the rare metal ore sorting device further includes:

[0133] The second dataset acquisition module is used to acquire a second ore dataset, wherein the second ore dataset is composed of ore images corresponding to different grades of ore, and the number of low-grade ore in the different grades of ore is higher than a preset quantity threshold, wherein the low-grade ore is ore with a target metal content lower than a preset content threshold.

[0134] The second detection module is used to detect ore sample images in the second ore dataset based on a pre-trained model to obtain a second detection result, wherein the second detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore.

[0135] The second parameter adjustment module is used to adjust the parameters in the pre-trained model based on the second detection result to obtain the ore sorting model.

[0136] In one embodiment, the rare metal ore sorting device further includes:

[0137] The third dataset acquisition module is used to acquire a third ore dataset, wherein the third ore dataset consists of ore sample images corresponding to the ore in the target mining area, and the ore in the target mining area consists of associated minerals and target metals in the target mining area;

[0138] The third detection module is used to detect the ore sample images in the third ore dataset based on the pre-trained model and obtain the third detection result, wherein the third detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore.

[0139] The third parameter adjustment module is used to adjust the parameters in the pre-trained model based on the third detection result to obtain the ore sorting model.

[0140] The rare metal ore sorting device provided in this application, employing the rare metal ore sorting method in the above embodiments, can solve the technical problem of low metal recovery rate in rare metal ore sorting. Compared with the prior art, the beneficial effects of the rare metal ore sorting device provided in this application are the same as those of the rare metal ore sorting method provided in the above embodiments, and other technical features in the rare metal ore sorting device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0141] This application provides a rare metal ore sorting device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the rare metal ore sorting method in the first embodiment described above.

[0142] The following is for reference. Figure 5 This document illustrates a structural schematic diagram of a rare metal ore sorting device suitable for implementing embodiments of this application. The rare metal ore sorting device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, tablets, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PMPs (Portable Media Players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5 The rare metal ore sorting equipment shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0143] like Figure 5 As shown, the rare metal ore sorting equipment may include a processing unit 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the rare metal ore sorting equipment. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the rare metal ore sorting equipment to exchange data wirelessly or via wired communication with other devices. Although rare metal ore sorting equipment with various systems is shown in the figure, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0144] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0145] The rare metal ore sorting equipment provided in this application, employing the rare metal ore sorting method described in the above embodiments, can solve the technical problem of low metal recovery rate in rare metal ore sorting. Compared with the prior art, the beneficial effects of the rare metal ore sorting equipment provided in this application are the same as those of the rare metal ore sorting method provided in the above embodiments, and other technical features of this rare metal ore sorting equipment are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0146] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0147] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0148] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the rare metal ore sorting method in the above embodiments.

[0149] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0150] The aforementioned computer-readable storage medium may be included in the rare metal ore sorting equipment; or it may exist independently and not assembled into the rare metal ore sorting equipment.

[0151] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the rare metal ore sorting equipment, cause the rare metal ore sorting equipment to perform the aforementioned rare metal ore sorting method.

[0152] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0153] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0154] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0155] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described rare metal ore sorting method, thereby solving the technical problem of low metal recovery rate in rare metal ore sorting. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the rare metal ore sorting method provided in the above embodiments, and will not be repeated here.

[0156] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for separating rare metal ores, characterized in that, The method includes: Obtain ore images; The ore corresponding to the ore image is sorted by a preset ore sorting model. The ore sorting model is trained based on rare metal ore sample images with different lithofacies characteristics. The rare metal ore is sorted based on the lithofacies characteristics of the ore. The lithofacies characteristics of the ore are the combination, structure and distribution of mineral components inside the ore. The ore sorting model has a channel attention module and a spatial attention module. The step of sorting the ore corresponding to the ore image using the preset ore sorting model includes: The channel attention module and the spatial attention module are used to extract features from the ore image to obtain the lithofacies features; Based on the lithofacies characteristics and the detection layer of the ore sorting model, a target detection result is obtained, wherein the target detection result includes whether the corresponding ore is a rare metal ore or not. Based on the target detection results, rare metal ores are sorted. The step of extracting features from the ore image using the channel attention module and the spatial attention module to obtain the lithofacies features includes: Based on the channel attention module, channel features are extracted from the ore image to obtain a channel feature map; Based on the spatial attention module, max pooling and average pooling are performed on the channel feature map to obtain max pooling features and average pooling features. The max pooling feature and the average pooling feature are concatenated to obtain the concatenation feature; Perform a convolution operation on the connection features to obtain the feature map spatial weights; The lithofacies features are obtained by dot product of the spatial weights of the feature map and the channel feature map.

2. The method as described in claim 1, characterized in that, Before the step of obtaining the ore image, the method further includes: Obtain a first ore dataset, wherein the first ore dataset consists of ore sample images corresponding to ores with radii smaller than a preset threshold; The ore sample images in the first ore dataset are detected based on the pre-trained model to obtain a first detection result, wherein the first detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore. Based on the first detection result, the preset parameters to be updated in the pre-trained model are adjusted to obtain the ore sorting model, wherein the parameters other than the parameters to be updated are frozen during the training process of the ore sorting model.

3. The method as described in claim 1, characterized in that, Following the step of obtaining the ore image, the method further includes: The ore image is preprocessed to obtain a preprocessed image; The preprocessed images of a preset number are stitched together to obtain the stitched image. The pre-set ore sorting model is used to sort the ore corresponding to the stitched image.

4. The method as described in claim 1, characterized in that, Before the step of obtaining the ore image, the method further includes: Obtain a second ore dataset, wherein the second ore dataset consists of ore images corresponding to ores of different grades, and among the ores of different grades, the number of low-grade ores is higher than a preset quantity threshold, and the low-grade ores are ores with a target metal content lower than a preset content threshold; The pre-trained model is used to detect the ore sample images in the second ore dataset to obtain a second detection result, wherein the second detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore. Based on the second detection result, the parameters in the pre-trained model are adjusted to obtain the ore sorting model.

5. The method as described in claim 1, characterized in that, Before the steps of obtaining the ore image, the method further includes: Obtain a third ore dataset, wherein the third ore dataset consists of ore sample images corresponding to the ore in the target mining area, and the ore in the target mining area consists of associated minerals and target metals in the target mining area; The ore sample images in the third ore dataset are detected based on the pre-trained model to obtain a third detection result, wherein the third detection result includes whether the corresponding ore is a rare metal ore or the corresponding ore is not a rare metal ore. Based on the third detection result, the parameters in the pre-trained model are adjusted to obtain the ore sorting model.

6. A rare metal ore sorting device, characterized in that, The device includes: The image acquisition module is used to acquire images of minerals; The ore sorting module is used to sort the ore corresponding to the ore image using a preset ore sorting model. The ore sorting model is trained based on rare metal ore sample images with different lithofacies characteristics. It sorts rare metal ores based on the lithofacies characteristics of the ore. The lithofacies characteristics of the ore are the combination, structure and distribution of mineral components inside the ore. The ore sorting model has a channel attention module and a spatial attention module. The ore sorting module further includes: The feature extraction submodule is used to extract features from the ore image through the channel attention module and the spatial attention module to obtain the lithofacies features; The ore detection submodule is used to perform detection based on the lithofacies characteristics and the detection layer of the ore sorting model to obtain the target detection result, wherein the target detection result includes whether the corresponding ore is a rare metal ore or whether the corresponding ore is not a rare metal ore. The ore sorting submodule is used to sort rare metal ores based on the target detection results; The feature extraction submodule further includes: The channel feature extraction unit is used to extract channel features from the ore image based on the channel attention module to obtain a channel feature map; The spatial feature extraction unit is used to perform max pooling and average pooling on the channel feature map based on the spatial attention module to obtain max pooling features and average pooling features. A feature connection unit is used to connect the max pooling feature and the average pooling feature to obtain a connection feature; The weight calculation unit is used to perform convolution operations on the connection features to obtain the feature map space weights; The feature dot product unit is used to perform a dot product between the spatial weights of the feature map and the channel feature map to obtain the lithofacies features.

7. A rare metal ore sorting device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the rare metal ore sorting method as described in any one of claims 1 to 5.

8. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the rare metal ore sorting method as described in any one of claims 1 to 5.

Images