Plant growth state feedback method and system based on image label

Abstract

The application relates to a plant growth state feedback method and system based on image labels, comprising the following steps: in response to a user query operation, obtaining plant label data uploaded by a target plant for the last time; receiving a natural language query request, analyzing a query intention and combining the plant label data to generate retrieval conditions; preferentially searching in a local database, and if the search fails, selecting a cross-unit search strategy according to a planting unit type; and generating feedback information according to a search result and displaying the feedback information to a user. The application reduces a query threshold through the combination of plant label data and natural language analysis, improves sample matching accuracy through the differentiation of two types of planting units, namely, planting cabins and outdoor planting areas, and the adoption of a differentiated cross-unit search strategy, realizes growth anomaly identification and intervention through timeliness judgment and multi-stage candidate set comparison, and enriches a sample library and improves system response efficiency through the localized storage of cross-unit search results.

Description

Technical Field

[0001] This application relates to the technical field of smart education, and in particular to a method and system for feedback on plant growth status based on image tags. Background Technology

[0002] As people pay increasing attention to healthy eating and green living, small-scale and decentralized planting scenarios such as home gardening, school science education planting, and community farms are becoming increasingly popular. Especially in the field of science education for primary and secondary schools, allowing students to participate in the plant planting process and observe the entire life cycle of plants from seed to maturity has become an important teaching method for cultivating observation skills, scientific literacy, and a sense of labor.

[0003] However, in actual planting and observation, users (especially primary and secondary school students) face a series of technical problems. Specifically, the long growth cycle of plants makes it difficult for users to continuously track morphological changes at each stage. When users want to understand the expected appearance of a target plant in subsequent stages, they often have to consult general books or online images, lacking real samples that match their own planting environment as a reference. On the other hand, different planting units (such as planting chambers and outdoor planting areas) have accumulated a large amount of historical observation data, but existing systems lack effective cross-unit sharing and retrieval mechanisms and fail to distinguish the differences in environmental dependence among different types of planting units, resulting in a large amount of valuable sample data not being fully utilized. In addition, it is difficult for users to establish an effective association between their query intent expressed through natural language and plant tag data, resulting in insufficient accuracy and relevance of search results.

[0004] In existing technologies, for example, CN117893914A discloses a plant growth monitoring method based on image recognition, including steps such as seed germination rate assessment, seedling root and hypocotyl analysis, leaf anomaly detection, flowering period and fruit fullness analysis, soil water separation mark analysis, and physiological activity monitoring, ultimately generating a plant growth monitoring report. However, this method only performs unidirectional image acquisition and analysis on plants within a single planting unit, and its monitoring results rely on the historical data accumulation of that unit itself. When local samples are insufficient, no reference basis can be obtained; moreover, this method only outputs growth data reports and cannot proactively provide intervention suggestions when plant growth deviates from expectations. Another example is CN121962716A, which discloses a plant growth stress monitoring and early warning system based on multispectral analysis. This system uses a multidimensional stress assessment module to calculate the direct stress intensity index, physiological rhythm disorder degree, and stress diffusion risk coefficient in parallel, and combines this with a three-dimensional early warning matrix to output graded early warning results. However, this scheme requires the collection of high-frequency multispectral image sequences at multiple time points within the complete diurnal cycle, and relies on a pre-constructed spectral stress standard knowledge base and standard physiological rhythm templates for comparative analysis, which places high demands on data acquisition equipment and template construction; moreover, its early warning results only output stress level and spread risk descriptions, and when the system lacks standard templates for specific stress types, it cannot provide real sample images for reference. Summary of the Invention

[0005] The purpose of this invention is to provide a plant growth status feedback method and system based on image tags, which partially solves or alleviates the above-mentioned shortcomings in the prior art, reduces the threshold for user query operations, improves the accuracy of cross-unit sample matching, realizes the identification and intervention of growth abnormalities, and continuously enriches the local sample library to improve the system response efficiency.

[0006] To solve the aforementioned technical problems, the present invention specifically adopts the following technical solution: A first aspect of the present invention is to provide a method for feeding back plant growth status based on image tags, characterized by comprising the following steps: Step S1: In response to the user's query operation for the target plant, retrieve the most recently uploaded plant tag data of the target plant from the local database of the current planting unit; The plant tag data includes at least the plant species, sample image, current growth stage, and health status. Step S2: Receive the user's natural language query request, parse and extract the query intent, combine the query intent with the plant tag data, and generate search conditions; Step S3: Perform a search in the local database based on the search criteria. If a search result matching the search criteria is found in the local database, proceed to step S6; otherwise, proceed to step S4. Step S4: Select the appropriate cross-unit retrieval strategy based on the cloud index information stored in the local database, and access the databases of other planting units in sequence according to the preset priority order for retrieval; The cloud-based index information includes identification information, type identification, geographical location, climate zone, establishment time, and sample richness. Step S5: If a search result matching the search criteria is found in the database of other planting units, proceed to step S6; If the search fails in the databases of other planting units, the visual language model is invoked to generate general feedback content, and the query ends. Step S6: Generate feedback information based on the search results and display it to the user, then end the query.

[0007] Furthermore, the types of planting units include a first type and a second type. The first type is a planting cabin equipped with a standardized environmental control module, and the second type is an outdoor planting area that depends on the natural environment.

[0008] Furthermore, if the current planting unit is of type 1, the cross-unit retrieval strategy in step S4 is specifically as follows: Retrieve all other first-type planting unit catalogs from the cloud, and obtain cloud index information for all planting units in the planting unit catalogs; Priority ranking results are obtained based on the establishment time of other first-type planting units, where the earlier the establishment time, the higher the priority ranking. According to the priority sorting results, the databases of each first-type planting unit are accessed sequentially for retrieval.

[0009] Furthermore, if the current planting unit is of the second type, the cross-unit retrieval strategy in step S4 is specifically as follows: Retrieve all other second-type planting unit directories from the cloud, and obtain cloud index information for all planting units in the planting unit directories; The other second-type planting units were weighted and ranked according to climate zone, geographical location, establishment time and sample richness to obtain priority ranking results; The databases of each second-type planting unit are accessed sequentially according to the priority sorting results.

[0010] Furthermore, the weighted scoring and ranking of other second-type planting units based on climate zone, geographical location, establishment time, and sample richness includes: The weight of samples with the same climate zone is the highest, followed by samples with geographical proximity. The weight of samples with the same climate zone is the second highest, while the weight of samples established and the richness of samples are tied for the lowest.

[0011] Furthermore, the query operation includes querying images of the target plant taken by the mobile terminal's camera.

[0012] Furthermore, the search results include at least one of the following: Historical growth stage sample images that match the search criteria; Plant label data corresponding to sample images from historical growth stages; Similarity score between sample images from historical growth stages and the current target plant.

[0013] Furthermore, the feedback method also includes the following steps: If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is less than or equal to a preset threshold, then the current growth stage of the target plant is determined based on the most recently uploaded plant tag data of the target plant. If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is greater than a preset threshold, the actual growth stage of the current target plant is determined based on the similarity score between the historical growth stage sample images and the current target plant.

[0014] Furthermore, the feedback method also includes the following steps: If a search result matching the search criteria is found in the database of another planting unit, the system will ask whether to copy and save the plant tag data retrieved across units to the local database. If the user agrees, the system will copy and save the data to the local database and update the cloud index information of the current planting unit.

[0015] A second aspect of the present invention is to provide a plant growth status feedback system based on image tags, the system comprising: The plant tag data acquisition module is configured to respond to a user's query operation for a target plant by retrieving the most recently uploaded plant tag data of the target plant from the local database of the current planting unit. The plant tag data includes at least the plant species, sample image, current growth stage, and health status. The search criteria generation module is configured to receive users' natural language query requests, parse and extract the query intent, combine the query intent with the plant tag data, and generate search criteria. The local retrieval module is configured to perform a retrieval in the local database based on the retrieval criteria. If a retrieval result matching the retrieval criteria is found in the local database, the retrieval feedback module's process is executed; otherwise, the cross-unit retrieval module's process is executed. The cross-unit retrieval module is configured to select the appropriate cross-unit retrieval strategy based on the cloud index information stored in the local database, and access the databases of other planting units in sequence according to the preset priority order for retrieval; The cloud-based index information includes identification information, type identification, geographical location, climate zone, establishment time, and sample richness. The retrieval judgment module is configured to execute the retrieval feedback module's process if a retrieval result matching the retrieval criteria is found in the database of other planting units. If the search fails in the databases of other planting units, the visual language model is invoked to generate general feedback content, and the query ends. The search feedback module is configured to generate feedback information based on the search results, display it to the user, and end the search.

[0016] Beneficial technical effects: 1. By combining plant tag data with natural language processing, the system significantly lowers the operational threshold for users and improves search accuracy. It automatically extracts the user's query intent from their natural language query request and combines it with the most recently uploaded plant tag data of the target plant to generate precise search criteria. Users do not need to manually input plant species or growth stages, making it especially suitable for primary and secondary school students with limited plant identification abilities.

[0017] 2. By classifying planting units by type and employing differentiated cross-unit retrieval strategies, the system achieves high-matching sample sharing and reuse. Based on the different environmental dependencies of the first type (planting cabin) and the second type (outdoor planting area), the system adopts priority ranking strategies based on either establishment time or a weighted score of climate zone, geographical location, establishment time, and sample richness. This ensures that the retrieved samples originate from other planting units with environmental conditions most similar to the current planting unit, thereby improving the reference value and transferability of cross-unit samples.

[0018] 3. By employing a weighted scoring and ranking system based on four dimensions—climate zone matching, geographical proximity, similar establishment time, and sample richness—the matching accuracy of cross-unit searches for the second type of outdoor planting areas was significantly improved. For outdoor planting areas dependent on the natural environment, the system uses a comprehensive weighted scoring and descending ranking based on similar climate zones, geographical proximity, similar establishment time, and sample richness. The system assigns the highest weight to areas with similar climate zones, followed by those with geographical proximity, while similar establishment time and sample richness have equal weight. This ensures that search results primarily originate from other outdoor planting areas with the most similar macro-environment and the richest data accumulation to the current planting unit. This effectively overcomes the impact of natural environmental differences on the comparability of plant growth and significantly improves the reference value of cross-unit samples.

[0019] 4. By locally storing cross-unit search results and dynamically updating the cloud index, the system continuously enriches the sample databases of each planting unit and improves system response efficiency. When a matching sample is found from another planting unit, the system asks the user whether to copy and save it to the local database. After confirmation, the system automatically stores and updates the total number of samples and sample richness fields in the cloud index information. This mechanism allows the local sample databases of each planting unit to continuously accumulate and improve with increased usage frequency, gradually reducing reliance on cross-unit searches, thereby reducing network access overhead and shortening the response time for subsequent similar queries. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. The elements or parts in the drawings are not necessarily drawn to scale. Obviously, the drawings described below are some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0021] Figure 1 This is a flowchart of a plant growth status feedback method based on image tags, as described in this application.

[0022] Figure 2 This is a schematic diagram of a planting unit in one embodiment of this application.

[0023] Figure 3 This is a schematic diagram of the module structure of a plant growth status feedback system based on image tags in one embodiment of this application.

[0024] Figure labeling: 101, Planting compartment; 102, Outdoor planting area; 201, Plant tag data acquisition module; 202, Search condition generation module; 203, Local search module; 204, Cross-unit search module; 205, Search judgment module; 206, Search feedback module. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0026] In this document, suffixes such as "module," "part," or "unit" used to denote elements are used only for the purpose of illustrative purposes and have no specific meaning in themselves. Therefore, "module," "part," or "unit" may be used interchangeably.

[0027] In this document, the terms "upper," "lower," "inner," "outer," "front," "rear," "one end," and "the other end," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the present invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0028] In this document, unless otherwise explicitly specified and limited, the terms "installed," "equipped with," "connected," etc., should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, a direct connection, or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0029] In this document, "and / or" includes any and all combinations of one or more of the listed related items.

[0030] In this article, "multiple" means two or more, that is, it includes two, three, four, five, etc.

[0031] Figure 1 A flowchart of a plant growth status feedback method based on image tags according to this application is shown. (Refer to...) Figure 1 The method specifically includes the following steps: Step S1: In response to the user's query operation for the target plant, retrieve the most recently uploaded plant tag data of the target plant from the local database of the current planting unit; The plant tag data includes at least the plant species, sample image, current growth stage, and health status. Step S2: Receive the user's natural language query request, parse and extract the query intent, combine the query intent with the plant tag data, and generate search conditions; Step S3: Perform a search in the local database based on the search criteria. If a search result matching the search criteria is found in the local database, proceed to step S6; otherwise, proceed to step S4. Step S4: Select the appropriate cross-unit retrieval strategy based on the cloud index information stored in the local database, and access the databases of other planting units in sequence according to the preset priority order for retrieval; The cloud-based index information includes identification information, type identification, geographical location, climate zone, establishment time, and sample richness. Step S5: If a search result matching the search criteria is found in the database of other planting units, proceed to step S6; If the search fails in the databases of other planting units, the visual language model is invoked to generate general feedback content, and the query ends. Step S6: Generate feedback information based on the search results and display it to the user, then end the query.

[0032] In some embodiments, in step S1, the query operation includes querying a current image of the target plant captured by a mobile terminal camera. In response to the user's query for the target plant, an existing image recognition model is used to identify the plant category in the current image, and the most recently uploaded plant tag data for the target plant is retrieved from the local database of the current planting unit. The plant tag data includes the plant species, sample image, current growth stage, and health status. In some embodiments, the plant tag data also includes data such as the time of tagging, the location of the shooting point, the season, weather information, ambient temperature, humidity, light conditions, and concurrent maintenance operation records.

[0033] In some embodiments, the type of planting unit includes a first type and a second type. For example... Figure 2 As shown, the first type is a planting chamber 101 equipped with a standardized environmental control module. This module includes a temperature control unit, a light control unit, a humidity control unit, and a nutrient solution supply unit, used to maintain the internal growth environment parameters of the planting chamber within the suitable growth range for the target plant, thereby decoupling plant growth from dependence on external geographical location and climate zone. The second type is an outdoor planting area 102 dependent on the natural environment, where the growth process of plants depends entirely on the natural climate conditions, soil characteristics, and sunlight, among other natural environmental factors, of its geographical location. By classifying planting units, this system can employ differentiated search strategies for different types of planting units during cross-unit search steps.

[0034] In some embodiments, if the current planting unit is of the first type, the cross-unit retrieval strategy in step S4 is specifically as follows: Retrieve all other first-type planting unit catalogs from the cloud, and retrieve the cloud index information of all planting units in the planting unit catalogs; Priority ranking results are obtained based on the establishment time of other first-type planting units, where the earlier the establishment time, the higher the priority ranking. According to the priority sorting results, the databases of each first-type planting unit are accessed sequentially for retrieval.

[0035] In a specific implementation, if the current planting unit is of the first type, the cross-unit retrieval strategy in step S4 is as follows: obtain the catalogs and corresponding cloud index information of all other first-type planting units from the cloud, and then sort them by priority based on the establishment time of each unit. Since the first-type planting cabin decouples its dependence on external geographical location and climate zone through the standardized environmental control module, the earlier the establishment time of the unit, the more complete its sample accumulation and the richer its time series data, so it is given a higher retrieval priority. The local databases of each first-type planting unit are accessed in order of establishment time from earliest to latest, and the retrieval stops immediately when a matching result is found for the first time, while recording the source planting unit identifier.

[0036] In some embodiments, if the current planting unit is of the second type, the cross-unit retrieval strategy in step S4 is specifically as follows: Retrieve all other second-type planting unit catalogs from the cloud, and retrieve the cloud index information for all planting units in the planting unit catalogs; The other second-type planting units were weighted and ranked according to climate zone, geographical location, establishment time and sample richness to obtain priority ranking results; The databases of each second-type planting unit are accessed sequentially according to the priority sorting results.

[0037] In some embodiments, weighted scoring and ranking of other second-type planting units based on climate zone, geographical location, establishment time, and sample richness includes: The weight of samples with the same climate zone is the highest, followed by samples with geographical proximity. The weight of samples with the same climate zone is the second highest, while the weight of samples established and the richness of samples are tied for the lowest.

[0038] In some embodiments, for planting units of the second type, the system obtains the catalog of all other second-type planting units from the cloud and obtains the cloud index information of each second-type planting unit.

[0039] Subsequently, based on four dimensions—climate zone, geographical location, establishment time, and sample richness—the other second-type planting units were ranked and weighted. For example, the weighting of each dimension followed the following principles: Climate Zone Dimension: If other second-type planting units are in the same climate zone as the current planting unit, this item scores 100 points; if they are in different climate zones, the similarity of the climate zones is further assessed: those belonging to adjacent climate zones (such as subtropical monsoon climate and tropical monsoon climate) score 50 points; those belonging to different major categories of climate zones (such as temperate continental climate and tropical rainforest climate) score 0 points. This dimension has a weight of 35%, and since climate zone is the most fundamental macro-environmental factor affecting plant growth, it is given the highest weight.

[0040] Geographical Location Dimension: If other second-type planting units are located within the same county-level administrative region as the current planting unit, this dimension scores 100 points; if located in different counties within the same prefecture-level city, the score is 70 points; if located in different prefecture-level cities within the same provincial-level administrative region, the score is 40 points; and if located in different provinces, the score is 10 points. This dimension has a weight of 25%. Geographical proximity implies more similar microclimate conditions, soil characteristics, and pest and disease prevalence patterns, resulting in stronger sample transferability.

[0041] Establishment Time Dimension: Second-type planting units established earlier have longer actual operating cycles, resulting in more complete and continuous plant growth sample data in terms of time sequence. This covers more growth cycles and richer environmental change scenarios, thus earning them a higher score. Specifically, all other second-type planting units are sorted according to their establishment time, with the earliest established unit receiving 100 points. The remaining units are assigned scores using linear normalization in reverse order of establishment time (i.e., earlier scores are higher), with the latest established unit receiving 0 points. This dimension has a weight of 20%.

[0042] Sample richness dimension: A normalized score is calculated based on the total number of valid sample images stored in the local database of the planting unit. Planting units with the top 20% of the total sample volume receive 100 points, those from 20% to 40% receive 80 points, those from 40% to 60% receive 60 points, those from 60% to 80% receive 40 points, and the bottom 20% receive 20 points. This dimension has a weight of 20%, and sample richness reflects the breadth of coverage and statistical reliability of the planting unit's database.

[0043] Finally, the weighted total score for each of the other second-type planting units is: climate zone score × 35% + geographical location score × 25% + establishment time score × 20% + sample richness score × 20%. The second-type planting units are then sorted in descending order of their weighted total scores, with higher scores indicating a closer match between the unit's samples and its actual growth conditions, thus giving it a higher search priority. Then, according to the sorting results, the databases of each second-type planting unit are accessed sequentially to check for matching search criteria.

[0044] In some embodiments, if the search fails in the databases of all other planting units, the system invokes an existing visual language model to generate general feedback content and terminates the query. When displaying this feedback, the system explicitly indicates that it originates from general knowledge generated by the visual language model, distinguishing it from feedback based on real samples. This fallback mechanism ensures that users receive feedback information under all circumstances.

[0045] In some embodiments, the search results include at least one of the following: Historical growth stage sample images that match the search criteria; Plant label data corresponding to sample images from historical growth stages; Similarity scores between historical growth stage sample images and the current target plant; In one specific implementation, the search results include at least one of the following: Historical growth stage sample images that match the search criteria. These sample images are real historical image data retrieved from a local database or a cross-institutional database that can satisfy the user's query intent. For example, when the user's query intent is to know the appearance of the next growth stage, the sample image is an example image of the next growth stage corresponding to the current target plant species and current growth stage; when the query intent is to obtain the maintenance record of the current stage, the sample image is a historical image that matches the current plant species and current growth stage, with accompanying maintenance operation tags (such as fertilization, watering, pruning, etc.).

[0046] This data includes plant tagging information for historical growth stage sample images. The tagging data comprises structured information generated during the historical annotation process, such as plant species, growth stage, health status, annotation timestamp, shooting season, and weather information. By providing this tagging data, users can understand the specific plant condition and growing environment at the time the sample image was taken, facilitating comparison and analysis with their own situation. For example, users can check whether the growth stage corresponding to the sample image is indeed the stage they expect to query, and whether the sample's health status was within the normal range at the time of shooting.

[0047] The similarity score is calculated between historical growth stage sample images and the current target plant. This similarity score is obtained by the system during the retrieval process using existing image feature extraction models for feature extraction and comparison. Specifically, models such as ResNet, ViT, or DINO can be used to extract image feature vectors from historical sample images and the current target plant image. Then, cosine similarity or Euclidean distance is calculated for feature comparison to measure the degree of similarity between the two in terms of plant morphology, leaf structure, color distribution, and texture features. The higher the calculated similarity score, the closer the visual state of the historical sample image is to the current target plant, and the greater its reference value to the user. When displaying the search results, the system can present this similarity score to the user in the form of a percentage, star rating, or progress bar to help the user quickly judge the reliability and applicability of the search results.

[0048] In some embodiments, the method of this application further includes the steps of: If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is less than or equal to a preset threshold, then the current growth stage of the target plant is determined based on the most recently uploaded plant tag data of the target plant. If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is greater than a preset threshold, the actual growth stage of the current target plant is determined based on the similarity score between the historical growth stage sample images and the current target plant.

[0049] In some embodiments, the method provided in this application further includes the step of determining the actual growth stage of the target plant based on a time interval and a similarity score: If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is less than or equal to a preset threshold, then the historical tag data is deemed to have sufficient timeliness to accurately reflect the current state of the target plant. In this case, the system directly determines the current growth stage of the target plant based on the growth stage field in the most recently uploaded plant tag data, and uses this as the basis for subsequent retrieval and feedback. For example, if the historical tag record shows that the target plant was in the "flowering stage" 3 days ago, and the preset threshold is 5 days, then the system directly determines that the target plant is still in the "flowering stage".

[0050] If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is greater than a preset threshold, the system determines that the historical tag data may not accurately reflect the actual state of the target plant due to the excessive time elapsed. In this case, the system does not directly use the growth stage in the historical tags, but instead determines the actual growth stage of the current target plant based on the similarity score between the historical growth stage sample images and the current target plant. Specifically, the system calculates the similarity score between the currently captured target plant image and the sample images of each historical growth stage in the database using an image similarity comparison model, and determines the growth stage corresponding to the sample image with the highest similarity score as the actual growth stage of the current target plant. For example, if the currently captured tomato image has the highest similarity score (e.g., 0.92) with the "fruiting stage" sample images in the database, but a lower similarity score (e.g., 0.45) with the "flowering stage" sample images, the system determines that the target plant has now entered the fruiting stage.

[0051] In some embodiments, the method of this application further includes the step of: If a search result matching the search criteria is found in the database of another planting unit, the system prompts the user to copy and save the sample images and plant tag data retrieved from the cross-unit search to the local database. If the user agrees, the copying and storage to the local database is executed, and the cloud index information of the current planting unit is updated. This mechanism gradually enriches the local sample database of the current planting unit during cross-unit searches, reducing the reliance of subsequent similar queries on cross-unit searches, thereby improving system response efficiency and reducing network access overhead.

[0052] In some embodiments, the method provided in this application further includes an anomaly detection and intervention step for the current growth stage: If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is less than or equal to a preset threshold, the system generates a set of candidate stages that the plant may currently be in based on the growth stage field in the historical tag data and the standard duration of that stage. For example, if the historical tag shows that the tomato was in the flowering and fruiting stage 5 days ago, and the standard duration of that stage is approximately 10 days, then 5 days ago it may have been at the beginning, middle, or end of the flowering and fruiting stage. Correspondingly, it may still be in the flowering and fruiting stage or has entered the fruit enlargement stage, thus generating a set of candidate stages {flowering and fruiting stage, fruit enlargement stage}. If the system finds no reference plant image for the fruit enlargement stage after searching the database of this unit or across units, then the flowering and fruiting stage is directly used as the only candidate expected stage. The system obtains the reference plant images corresponding to each stage in the candidate stage set from the database of this unit or across units, and compares the similarity of the currently captured target plant image with the reference images of each candidate stage one by one. If the highest similarity score is greater than or equal to the preset matching threshold (e.g., 0.85), the plant is considered to be growing normally and is consistent with the actual stage. If the similarity scores of all candidate stages are lower than the threshold (e.g., the highest score is only 0.58, and the scores of all candidate stages are below 0.85), the plant is considered to be growing abnormally and its morphological development deviates from the characteristics of any standard stage. At this time, the system triggers the abnormal handling process, searches for similar cases in the local database or the database of other units, and provides intervention suggestions. If no similar cases are found, the system obtains the environmental parameter monitoring data of the current planting unit through the planting chamber and compares each parameter such as temperature, light, humidity, and nutrient solution to see if they are within the suitable range for the plant in the current growth stage. If a parameter is not within the suitable range for the current growth stage, the parameter is identified as an abnormal parameter, and the system attempts to adjust the parameter to the suitable range for the current growth stage through the planting chamber. If the adjustment cannot be performed normally, the system prompts for manual intervention for maintenance.

[0053] In some embodiments, if the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is greater than a preset threshold, the system first obtains typical sample images of the target plant species at each standard growth stage from the database of the unit or across units, and simultaneously obtains the standard reference days between each standard growth stage. The standard growth stages are configured according to different plant species. For example, for tomatoes, they can be divided into germination stage, seedling stage, flowering and fruit setting stage, fruit enlargement stage, green ripening stage, and ripening stage. The standard reference days between each stage are, for example, 12 days from flowering and fruit setting stage to fruit enlargement stage, 18 days from fruit enlargement stage to green ripening stage, and 6 days from green ripening stage to ripening stage.

[0054] Then, based on the historical growth stages in the most recently uploaded plant tag data of the target plant, the time interval between the timestamp of the historical tag data and the current query operation, and the standard reference days between each stage, the expected set of growth stages that the target plant may currently be in is calculated. Since the time recorded by the historical tag may be at different locations such as the beginning, middle, or end of the historical growth stage, this application uses a time window expansion method, using the standard duration of the historical growth stage as the uncertainty interval, and compares the time interval with the cumulative days of each stage to obtain one or more possible candidate stages, for example: If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant exceeds a preset threshold, the system first retrieves typical sample images of the target plant species at each standard growth stage from its own or cross-institutional database, and simultaneously obtains the standard reference number of days between each standard growth stage. Specifically, if historical tags show that the tomato was in the flowering and fruiting stage 30 days ago, the standard duration of that flowering and fruiting stage is approximately 10 days. Since the actual growth progress 30 days ago may have been at the beginning, middle, or end of the flowering and fruiting stage, the current stage after 30 days will present multiple possibilities. If the historical record corresponds to the beginning of the flowering and fruit-setting period, then after 12 days it enters the fruit enlargement stage, and after another 18 days it enters the green-ripe stage, accumulating 30 days to reach the green-ripe stage. If the historical record corresponds to the midpoint of the flowering and fruit-setting period (approximately day 5), then after the remaining 5 days it leaves the flowering and fruit-setting period, then after 12 days of fruit enlargement and 13 days of green-ripe stage, accumulating 30 days, it should be in the late green-ripe stage. If the historical record corresponds to the end of the flowering and fruit-setting period (i.e., about to enter the fruit enlargement stage), then after 30 days it should be after the green-ripe stage, close to the ripening stage. Thus, a candidate stage set is generated, for example, {green-ripe stage, ripening stage}. If a search of the unit's or cross-unit databases reveals that no corresponding reference plant images exist for any stage in the candidate stage set, i.e., no sample images of the green-ripe stage or ripening stage are found, the system cannot make an anomaly judgment through image comparison. In this case, a "Cannot determine an anomaly" message is displayed, and manual intervention for on-site observation and diagnosis is recommended. If at least one reference plant image corresponding to a candidate stage is retrieved, the system will compare the similarity of the currently captured target plant image with the reference images of each candidate stage one by one, and determine whether the current plant growth is normal or abnormal based on the comparison results.

[0055] After determining the candidate stage set, the system retrieves reference plant images corresponding to each candidate stage from the database and compares the similarity of the currently captured target plant image with the reference image of each candidate stage one by one. The system determines the candidate stage with the highest similarity score as the actual growth stage of the current target plant.

[0056] If the highest similarity score is greater than or equal to the preset matching threshold (e.g., 0.85), the plant is considered to be growing normally and consistent with its actual stage. If the similarity scores of all candidate stages are lower than the threshold (e.g., the highest score is only 0.62, and none of the candidate stage scores reach 0.85), the plant is considered to be growing abnormally, and its morphological development deviates from the characteristics of any standard stage. In response to the anomaly determination, the system further searches the local database and cross-unit database for similar cases that match the current plant species and the current anomaly type, and obtains the intervention methods used in the case and the recovery effect after intervention as suggestions to be fed back to the user. If no similar cases are found, the system obtains the environmental parameter monitoring data of the current planting unit through the planting chamber, and compares each parameter such as temperature, light, humidity, and nutrient solution to see if they are within the suitable range for the plant in the current growth stage. If a parameter is not within the suitable range for the current growth stage, it is identified as an abnormal parameter, and the system attempts to adjust the parameter to the suitable range for the current growth stage through the planting chamber. If it cannot be adjusted normally, the system prompts for manual intervention for maintenance.

[0057] Further reference Figure 3 As an implementation of the above-described method, this application provides an embodiment of a plant growth status feedback system based on image tags. This system embodiment is similar to... Figure 1 Corresponding to the method embodiments shown, the system can be specifically applied to various electronic devices.

[0058] refer to Figure 3 A plant growth status feedback system based on image tags, comprising: The plant tag data acquisition module 201 is configured to, in response to a user's query operation for a target plant, acquire the most recently uploaded plant tag data of the target plant from the local database of the current planting unit; The plant tag data includes at least the plant species, sample image, current growth stage, and health status. The search condition generation module 202 is configured to receive a user's natural language query request, parse and extract the query intent, combine the query intent with the plant tag data, and generate search conditions. The local retrieval module 203 is configured to perform a retrieval in the local database based on the retrieval conditions. If a retrieval result matching the retrieval conditions is found in the local database, the retrieval feedback module process is executed; otherwise, the cross-unit retrieval module process is executed. The cross-unit retrieval module 204 is configured to select the appropriate cross-unit retrieval strategy based on the cloud index information stored in the local database, and access the databases of other planting units in sequence according to the preset priority order for retrieval; The cloud-based index information includes identification information, type identification, geographical location, climate zone, establishment time, and sample richness. The retrieval judgment module 205 is configured to execute the process of the retrieval feedback module if a retrieval result matching the retrieval conditions is found in the database of other planting units. If the search fails in the databases of other planting units, the visual language model is invoked to generate general feedback content, and the query ends. The search feedback module 206 is configured to generate feedback information based on the search results and display it to the user, and then end the query.

[0059] In another aspect, this application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the following... Figure 1 The method shown.

[0060] 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 apparatus 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 apparatus. 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 apparatus that includes that element.

[0061] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a computer terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0062] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for feeding back plant growth status based on image tags, characterized in that, Includes the following steps: Step S1: In response to the user's query operation for the target plant, retrieve the most recently uploaded plant tag data of the target plant from the local database of the current planting unit; The plant tag data includes at least the plant species, sample image, current growth stage, and health status. Step S2: Receive the user's natural language query request, parse and extract the query intent, combine the query intent with the plant tag data, and generate search conditions; Step S3: Perform a search in the local database based on the search criteria. If a search result matching the search criteria is found in the local database, proceed to step S6; otherwise, proceed to step S4. Step S4: Select the appropriate cross-unit retrieval strategy based on the cloud index information stored in the local database, and access the databases of other planting units in sequence according to the preset priority order for retrieval; The cloud-based index information includes identification information, type identification, geographical location, climate zone, establishment time, and sample richness. Step S5: If a search result matching the search criteria is found in the database of other planting units, proceed to step S6; If the search fails in the databases of other planting units, the visual language model is invoked to generate general feedback content, and the query ends. Step S6: Generate feedback information based on the search results and display it to the user, then end the query.

2. The plant growth status feedback method based on image tags according to claim 1, characterized in that: The types of planting units include a first type and a second type. The first type is a planting cabin equipped with a standardized environmental control module, and the second type is an outdoor planting area that depends on the natural environment.

3. The method for plant growth status feedback based on image tags according to claim 2, characterized in that: If the current planting unit is of type 1, the cross-unit retrieval strategy in step S4 is specifically as follows: Retrieve all other first-type planting unit catalogs from the cloud, and obtain cloud index information for all planting units in the planting unit catalogs; Priority ranking results are obtained based on the establishment time of other first-type planting units, where the earlier the establishment time, the higher the priority ranking. According to the priority sorting results, the databases of each first-type planting unit are accessed sequentially for retrieval.

4. The plant growth status feedback method based on image tags according to claim 2, characterized in that: If the current planting unit is of type two, the cross-unit retrieval strategy in step S4 is specifically as follows: Retrieve all other second-type planting unit directories from the cloud, and obtain cloud index information for all planting units in the planting unit directories; The other second-type planting units were weighted and ranked according to climate zone, geographical location, establishment time and sample richness to obtain priority ranking results; The databases of each second-type planting unit are accessed sequentially according to the priority sorting results.

5. The plant growth status feedback method based on image tags according to claim 4, characterized in that: The weighted scoring and ranking of other second-type planting units based on climate zone, geographical location, establishment time, and sample richness includes: The weight of samples with the same climate zone is the highest, followed by samples with geographical proximity. The weight of samples with the same climate zone is the second highest, while the weight of samples established and the richness of samples are tied for the lowest.

6. The plant growth status feedback method based on image tags according to claim 1, characterized in that: The query operation includes querying images of the target plant taken by the mobile terminal's camera.

7. The plant growth status feedback method based on image tags according to claim 6, characterized in that: The search results include at least one of the following: Historical growth stage sample images that match the search criteria; Plant label data corresponding to sample images from historical growth stages; Similarity score between sample images from historical growth stages and the current target plant.

8. The plant growth status feedback method based on image tags according to claim 7, characterized in that, It also includes the following steps: If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is less than or equal to a preset threshold, then the current growth stage of the target plant is determined based on the most recently uploaded plant tag data of the target plant. If the time interval between the most recently uploaded plant tag data of the target plant and the current query operation on the target plant is greater than a preset threshold, the actual growth stage of the current target plant is determined based on the similarity score between the historical growth stage sample images and the current target plant.

9. A method for feeding back plant growth status based on image tags according to any one of claims 1-8, characterized in that, It also includes the following steps: If a search result matching the search criteria is found in the database of another planting unit, the system will ask whether to copy and save the plant tag data retrieved across units to the local database. If the user agrees, the system will copy and save the data to the local database and update the cloud index information of the current planting unit.

10. A plant growth status feedback system based on image tags, characterized in that, The system includes: The plant tag data acquisition module is configured to respond to a user's query operation for a target plant by retrieving the most recently uploaded plant tag data of the target plant from the local database of the current planting unit. The plant tag data includes at least the plant species, sample image, current growth stage, and health status. The search criteria generation module is configured to receive users' natural language query requests, parse and extract the query intent, combine the query intent with the plant tag data, and generate search criteria. The local retrieval module is configured to perform a retrieval in the local database based on the retrieval criteria. If a retrieval result matching the retrieval criteria is found in the local database, the retrieval feedback module's process is executed; otherwise, the cross-unit retrieval module's process is executed. The cross-unit retrieval module is configured to select the appropriate cross-unit retrieval strategy based on the cloud index information stored in the local database, and access the databases of other planting units in sequence according to the preset priority order for retrieval; The cloud-based index information includes identification information, type identification, geographical location, climate zone, establishment time, and sample richness. The retrieval judgment module is configured to execute the retrieval feedback module's process if a retrieval result matching the retrieval criteria is found in the database of other planting units. If the search fails in the databases of other planting units, the visual language model is invoked to generate general feedback content, and the query ends. The search feedback module is configured to generate feedback information based on the search results, display it to the user, and end the search.

Images