1. A crop disease recognition system (100), comprising: at least one processor; and a memory storing instructions that, when executed by the at least one processor, cause the system to: receive one or more images depicting crop (102) plant tissue; compute an image quality assessment (106) based on one or more quality metrics; preprocess at least one accepted image based on the image quality score; segmentation (108) of the plant tissue and/or lesion regions from the at least one accepted image; a trained machine learning model (110) to the segmented output (116) to generate a disease classification and a confidence score; and wherein the output (116) disease classification and the confidence score through a user interface.

2. The system (100) as claimed in claim 1, wherein the system is configured to output (116) a severity estimate based on a lesion-area ratio derived from segmentation (108).

3. The system (100) as claimed in claim 1, wherein the system provides capture guidance to a user when the image quality assessment (106) score meets a rejection condition.

4. The system (100) as claimed in claim 1, wherein the receiving one or more images includes receiving multiple images of a same plant region, and the system generates a fused disease segmentation (108) based on the images.

5. The system (100) as claimed in claim 4, wherein the fusion is performed using confidence-weighted averaging of class probabilities.

6. The system (100) as claimed in claim 1, comprising a hybrid deployment controller (112) configured to perform initial inference on an edge device and selectively transmit data to a cloud (114) system based on confidence level and connectivity.

7. The system (100) as claimed in claim 6, wherein the cloud (114) system uses an ensemble of models to generate an output (116) to generate updated disease classification and confidence score.

8. The system (100) as claimed in claim 1, wherein the image quality assessment (106) metrics include blur, illumination adequacy, and plant coverage.

9. The system (100) as claimed in claim 1, wherein preprocessing includes color normalization and illumination correction.

10. The system (100) as claimed in claim 1, comprising storing diagnosis history with location and time information and presenting a trend report.  

6. DATE AND SIGNATURE Dated this on 09th April, 2026 Signature Mr. Srinivas Maddipati (IN/PA 3124) Agent for applicant

Description:4. DESCRIPTION
Technical Field of the Invention

The present invention relates to agricultural decision support systems. More particularly, the invention relates to computer-implemented systems for detecting, classifying, and assessing crop diseases using image analysis and machine learning, including hybrid edge/cloud deployment for robust performance under real-world field conditions.
Background of the Invention

Crop diseases adversely affect agricultural productivity by reducing both yield and quality, and may spread rapidly if not detected at an early stage. Conventional disease detection techniques primarily rely on visual scouting by farmers or agronomists, which is time-consuming, subjective, and highly dependent on the expertise of the observer.

In recent years, image-based crop disease recognition using Machine Learning techniques has emerged as an alternative approach. However, existing solutions generally perform effectively only under controlled conditions, such as uniform backgrounds, consistent lighting, and high-quality images.

In practical agricultural environments, the performance of such systems degrades significantly due to challenges including variable illumination, occlusions, complex and mixed backgrounds, camera motion blur, variations in crop cultivars, and similarities between disease symptoms and other plant stresses such as nutrient deficiencies.

Real-world deployment of such systems often requires operation in regions with limited or intermittent network connectivity, thereby necessitating efficient on-device inference capabilities while still enabling integration with cloud-based resources for computationally intensive processing, model updates, and expert-level analysis.

Accordingly, there exists a need for an integrated and robust system capable of addressing the aforementioned limitations by evaluating and enhancing image quality, supporting both single-image and multi-image diagnosis, providing disease severity estimation along with actionable recommendations, and continuously improving performance through feedback-driven adaptation and retraining, while effectively managing constraints associated with edge and cloud environments.

Objects of the Invention

The principal object of the present invention to provide a system for recognizing crop diseases from images in a reliable manner.

Another object of the invention is to improve image quality before analysis so that detection is accurate even in real field conditions.

Another object of the invention is to identify diseases from one or more images of crops.

Another object of the invention is to estimate the severity of the disease and provide useful recommendations.

Another object of the invention is to work in low network areas using on-device processing along with optional cloud support.

Another object of the invention is to adapt and improve over time using feedback and updated data.

Another object of the invention is to provide a system that works consistently under different lighting, background, and environmental conditions.

Brief Summary of the Invention

The present invention relates to a system for recognizing crop diseases using images in a reliable and practical manner. The system receives one or more images of a crop area captured through a user device such as a smartphone or any imaging platform designed to work effectively in real field conditions rather than controlled environments.

According to an aspect of the present invention, a robust image-based system includes a quality assessment step that checks whether the captured image is suitable for analysis. It evaluates factors such as blur, lighting, and coverage of the crop area, and provides a quality score. Based on this, the system may accept the image, reject it, or guide the user to capture a better image.

In another aspect of the present invention, the images are then processed using a preprocessing module that improves consistency across different devices and environments. This includes adjusting lighting, reducing noise, enhancing contrast, and resizing the image. A segmentation step may further isolate the plant or affected region to reduce background interference.

In another aspect of the present invention, an inference engine using trained machine learning models analyses the processed image to identify the disease. The system provides outputs such as disease type, confidence level, and severity of the disease. It may also indicate similar conditions, such as nutrient deficiencies, to avoid misclassification.

In another aspect of the present invention, a hybrid approach where initial analysis is performed on the device, and additional processing can be carried out using cloud resources when required. This allows the system to function even in low network conditions while still benefiting from advanced analysis when connectivity is available.

In another aspect of the present invention, the system provides recommendations such as treatment options, monitoring steps, and alerts. It maintains records with location and time details, and includes a feedback mechanism where user inputs or expert validation are used to improve the model over time through retraining and updates.

Brief Description of the Drawings

The invention will be further understood from the following detailed description of a preferred embodiment taken in conjunction with an appended drawing, in which:

Fig. 1 illustrates a system (100) for robust image-based crop disease recognition, in accordance with an exemplary embodiment of the present invention;

Detailed Description of the Invention

It is to be understood that the present disclosure is not limited in its application to the details of construction and the arrangement of components set forth in the following description or illustrated in the drawings. The present disclosure is capable of other embodiments and of being practiced or of being carried out in various ways. In addition, it is to be understood that the phraseology and terminology used herein is for the purpose of description and should not be regarded as limiting.

The use of “including”, “comprising” or “having” and variations thereof herein is meant to encompass the items listed thereafter and equivalents thereof as well as additional items. The terms “a” and “an” herein do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced items. Further, the use of terms “first”, “second”, and “third”, and the like, herein do not denote any order, quantity, or importance, but rather are used to distinguish one element from another.

According to an exemplary embodiment of the present invention, a system for recognizing crop diseases from images in a reliable manner under field conditions. The system is implemented using one or more processors on an edge device and/or cloud platform, and is designed to handle variations such as lighting, background, device differences, and image quality.

In accordance with an exemplary embodiment of the present invention, wherein the user device configured to capture one or more images of crop plant tissue. The device may be a smartphone or similar imaging device, and may also collect additional information such as time, location, and crop type to support analysis.

In accordance with an exemplary embodiment of the present invention, wherein the quality of the captured images before processing. The system determines whether the image is suitable by analyzing parameters such as blur, lighting, and coverage. If the image is not suitable, the system provides guidance to the user to capture a better image.

In accordance with an exemplary embodiment of the present invention, wherein the system processes accepted images using preprocessing techniques. These include resizing, color correction, noise reduction, contrast enhancement, and illumination adjustment to ensure consistent input for further analysis.

In accordance with an exemplary embodiment of the present invention, wherein the system isolates the plant region from the background and identifies relevant areas such as affected regions. This step helps reduce interference from surrounding elements and improves accuracy of disease detection.

In accordance with an exemplary embodiment of the present invention, wherein the processed image is analyzed using one or more trained machine learning models. The system identifies the type of disease and provides a confidence level for the prediction. In certain cases, the system may also indicate uncertainty or similar conditions.

In accordance with an exemplary embodiment of the present invention, wherein the multiple images of the same crop are provided, the system combines the results to generate a more reliable output. This may involve combining predictions based on confidence or excluding low-quality inputs.

In accordance with an exemplary embodiment of the present invention, wherein the severity of the disease based on the extent of the affected area. The severity may be expressed as a percentage or categorized into levels such as mild, moderate, or severe.

In accordance with an exemplary embodiment of the present invention, wherein the system supports operation in both device-based and cloud-based environments. Initial analysis may be performed on the device, and if required, additional processing may be carried out using cloud resources to improve accuracy.

In accordance with an exemplary embodiment of the present invention, wherein the provided outputs include disease identification, confidence level, severity, and recommended actions. The system may also store records with time and location information for tracking and reporting purposes.

In accordance with an exemplary embodiment of the present invention, wherein a feedback mechanism through which users or experts can provide input on the correctness of the results. This feedback is used to improve the system over time through updates and retraining.

In reference to the figures, fig. 1 illustrates a system (100) for robust image-based crop (102) disease recognition. The system receives crop images through an image acquisition (104) module and evaluates their suitability using a quality assessment (106) module. Accepted images are processed in a preprocessing module for enhancement and normalization, and then provided to a segmentation (108) module to isolate plant regions. The segmented output is analyzed by a machine learning model (110) to identify diseases and generate outputs (116) including disease label, confidence score, and severity. The system may use a hybrid controller (112) for edge and cloud (114) processing, and the results are displayed through an output module.
, Claims:5. CLAIMS
I/We Claim:
1. A crop disease recognition system (100), comprising:
at least one processor; and
a memory storing instructions that, when executed by the at least one processor, cause the system to:
receive one or more images depicting crop (102) plant tissue;
compute an image quality assessment (106) based on one or more quality metrics;
preprocess at least one accepted image based on the image quality score;
segmentation (108) of the plant tissue and/or lesion regions from the at least one accepted image;
a trained machine learning model (110) to the segmented output (116) to generate a disease classification and a confidence score; and
wherein the output (116) disease classification and the confidence score through a user interface.

2. The system (100) as claimed in claim 1, wherein the system is configured to output (116) a severity estimate based on a lesion-area ratio derived from segmentation (108).

3. The system (100) as claimed in claim 1, wherein the system provides capture guidance to a user when the image quality assessment (106) score meets a rejection condition.

4. The system (100) as claimed in claim 1, wherein the receiving one or more images includes receiving multiple images of a same plant region, and the system generates a fused disease segmentation (108) based on the images.
5. The system (100) as claimed in claim 4, wherein the fusion is performed using confidence-weighted averaging of class probabilities.

6. The system (100) as claimed in claim 1, comprising a hybrid deployment controller (112) configured to perform initial inference on an edge device and selectively transmit data to a cloud (114) system based on confidence level and connectivity.

7. The system (100) as claimed in claim 6, wherein the cloud (114) system uses an ensemble of models to generate an output (116) to generate updated disease classification and confidence score.

8. The system (100) as claimed in claim 1, wherein the image quality assessment (106) metrics include blur, illumination adequacy, and plant coverage.

9. The system (100) as claimed in claim 1, wherein preprocessing includes color normalization and illumination correction.

10. The system (100) as claimed in claim 1, comprising storing diagnosis history with location and time information and presenting a trend report.

 
6. DATE AND SIGNATURE
Dated this on 09th April, 2026
Signature

Mr. Srinivas Maddipati
(IN/PA 3124)
Agent for applicant