1. A method for designing and implementing a baseline Liquid Neural Network (108) architecture for multispectral Land Use and Land Cover (LULC) classification, a. multispectral satellite data (102) comprising multiple spectral bands is collected for representing different land surface characteristics; b. the collected multispectral satellite data (102) is preprocessed (104) by performing normalization and noise removal to improve data quality; c. features are extracted (106) from the preprocessed (104) multispectral satellite data (102) to represent land characteristics; d. a Liquid Neural Network (108) architecture is designed comprising adaptive neurons configured to process time-varying input data; e. the Liquid Neural Network (108) architecture is implemented and trained using the extracted features to learn patterns associated with different land types; f. the multispectral satellite data (102) is classified into one or more land categories (110) including forest, water, urban, and agricultural land; and g. a classification output (112) is generated representing identified land use and land cover categories.

2. The method as claimed in claim 1, wherein the multispectral satellite data (102) comprises visible, near-infrared, and shortwave infrared spectral bands.

3. The method as claimed in claim 1, wherein the preprocessing (102) step comprises data normalization, noise filtering, and removal of inconsistencies in the multispectral data.

4. The method as claimed in claim 1, wherein the feature extraction (104) step comprises extraction of spatial and spectral features representing land characteristics.

5. The method as claimed in claim 1, wherein the Liquid Neural Network (106) architecture comprises adaptive neurons having time-varying parameters for processing dynamic input data.

6. The method as claimed in claim 1, wherein the training of the Liquid Neural Network (106) is performed using labeled multispectral datasets representing different land categories.

7. The method as claimed in claim 1, wherein the classification step includes land categorizing (110) into forest, water bodies, urban areas, and agricultural regions.

8. The method as claimed in claim 1, wherein the classification output (112) is presented in the form of maps, reports, or visual representations.

9. The method as claimed in claim 1, wherein the method is configured to handle variations in environmental conditions including seasonal changes, illumination differences, and noise in satellite imagery.

10. The method as claimed in claim 1, wherein the method provides a baseline implementation capable of optimization for improved accuracy and scalability.  

6. DATE AND SIGNATURE Dated this on 06th 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 the field of remote sensing and machine learning. More particularly, the invention relates to a method for designing and implementing a Liquid Neural Network architecture for multispectral land use and land cover (LULC) classification.

Background of the Invention

Land Use and Land Cover (LULC) classification is important for applications such as environmental monitoring, urban planning, and resource management. Multispectral satellite images are widely used as they provide detailed information across different spectral bands, enabling identification of various land types.

Conventional methods such as rule-based systems and traditional machine learning approaches are commonly used for LULC classification. However, these methods have limitations in handling complex patterns and variations present in multispectral data, resulting in reduced accuracy.

Deep learning models, including convolutional neural networks, have improved classification performance. Despite this, they require large datasets and high computational resources, making them less efficient in many practical situations.

Adaptive neural network models have been introduced to handle dynamic and complex data more effectively. However, their application in multispectral LULC classification is still limited, and simple baseline implementations are not widely available.

Existing methods often fail to fully utilize multispectral data and adapt to changing environmental conditions, leading to inconsistent results.
Accordingly, there exists a need for a method that uses adaptive neural network architectures for accurate and efficient multispectral LULC classification. The present invention addresses these limitations by providing a baseline Liquid Neural Network based method for reliable land classification.

Objects of the Invention

The principle object of the present invention is to provide a method for designing and implementing a baseline Liquid Neural Network architecture for multispectral LULC classification.

Another object of the present invention is to improve classification accuracy by utilizing adaptive learning capabilities of the Liquid Neural Network.

Another object of the present invention is to effectively process multispectral satellite data for identifying different land categories.

Another object of the present invention is to provide a baseline model that can be easily implemented and further optimized for improved performance.

Another object of the present invention is to handle complex patterns and variations present in multispectral data using adaptive neural network techniques.

Another object of the present invention is to reduce dependency on large datasets and high computational resources compared to conventional deep learning models.

Another object of the present invention is to enhance the capability of classification under varying environmental conditions such as seasonal and spectral variations.

Another object of the present invention is to provide a reliable and efficient method for real world applications including environmental monitoring, urban planning, and agricultural analysis.

Brief Summary of the Invention

The present invention provides a method for designing and implementing a baseline Liquid Neural Network architecture for multispectral Land Use and Land Cover (LULC) classification. The method focuses on improving the way satellite images are analyzed for identifying different land types.

The invention uses multispectral satellite data, which contains information from multiple spectral bands. This helps in capturing more detailed characteristics of land surfaces compared to conventional methods.

The method includes preprocessing of input data to improve quality and remove inconsistencies. This involves operations such as normalization, noise removal, and preparation of data to ensure that variations in the input do not affect the performance of the model and that the data is suitable for accurate classification.

A Liquid Neural Network model is designed and implemented to process the multispectral data. The model has adaptive properties, allowing it to handle complex patterns and variations effectively.

The model is trained using the processed data to learn different land characteristics. Based on this learning, it classifies the input data into categories such as forest, water, urban, and agricultural land.

The invention provides a baseline implementation that can be further improved for better performance. It also reduces dependency on large datasets and high computational requirements.
The method ensures more accurate and reliable classification results, even under varying environmental conditions such as seasonal changes, lighting differences, and noise in satellite images, thereby improving robustness of the system.

The invention offers a simple, efficient, and adaptable approach for multispectral LULC (102) classification, which can be easily implemented and extended for real-world applications including environmental monitoring, urban planning, and resource

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 flow diagram of a method for multispectral land use and land cover (LULC) classification using a Liquid Neural Network (108) architecture, 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 method is provided for designing and implementing a baseline Liquid Neural Network architecture for multispectral Land Use and Land Cover (LULC) classification. The method focuses on improving classification accuracy by using adaptive learning techniques. It provides a structured approach for processing satellite data and generating reliable results. The invention is suitable for handling complex and dynamic datasets.

In accordance with an exemplary embodiment of the present invention, multispectral satellite data is collected, wherein the data comprises multiple spectral bands capturing different characteristics of land surfaces. These bands may include visible, near-infrared, and other spectral ranges. The use of multispectral data enables better differentiation between land types. This improves the overall effectiveness of classification.

In accordance with an exemplary embodiment of the present invention, the collected data is subjected to pre-processing, which includes normalization and noise removal. These steps help in improving data quality and removing unwanted variations. Pre-processing ensures that the data is consistent and suitable for further analysis. This step plays an important role in enhancing model performance.

In accordance with an exemplary embodiment of the present invention, features are extracted from the pre-processed multispectral data to represent spatial and spectral characteristics of land regions. Feature extraction helps in identifying meaningful patterns in the data. These features act as inputs to the neural network for classification. Proper feature representation improves accuracy and reliability.
In accordance with an exemplary embodiment of the present invention, a baseline Liquid Neural Network architecture is designed, wherein the network comprises adaptive neurons capable of processing time-varying input data. The architecture is flexible and can adjust based on input variations. This makes it suitable for handling complex and dynamic environments. The design forms the core of the invention.

In accordance with an exemplary embodiment of the present invention, the designed Liquid Neural Network is implemented and trained using the extracted features. During training, the model learns patterns associated with different land types. The adaptive nature of the network allows it to capture complex relationships in the data. This results in improved classification capability.

In accordance with an exemplary embodiment of the present invention, the trained model processes the multispectral data and classifies it into categories such as forest, water, urban, and agricultural land. The classification is performed based on learned patterns and features. This step provides meaningful insights into land usage. It supports various practical applications.

In accordance with an exemplary embodiment of the present invention, the classification results are generated as output data representing identified land use and land cover categories. The output may be presented in the form of maps, reports, or visual representations. This makes it easier for users to interpret the results. It also supports decision-making processes.

In accordance with an exemplary embodiment of the present invention, the method is capable of handling variations in multispectral data, including changes in environmental conditions, lighting, and noise. The adaptive nature of the Liquid Neural Network improves robustness. It ensures consistent performance under different scenarios. This is a key advantage of the invention.

In accordance with an exemplary embodiment of the present invention, the baseline model provides a simple and effective implementation that can be further enhanced. It serves as a starting point for developing advanced models. Additional improvements can be made for higher accuracy and efficiency. This makes the invention flexible and scalable.

In accordance with an exemplary embodiment of the present invention, the method ensures efficient processing of multispectral data while reducing dependency on large datasets and high computational resources. This makes it practical for real-world applications. It also reduces implementation complexity. The method balances performance and efficiency.

In accordance with an exemplary embodiment of the present invention, the invention provides a reliable and adaptable solution for multispectral LULC classification. It can be applied in environmental monitoring, urban planning, agriculture, and resource management. The method supports accurate and efficient land analysis. It contributes to improved decision-making and planning.

In references to figures, Fig. 1 illustrates a flow diagram of a method for classifying multispectral satellite data (102) using a Liquid Neural Network. The process begins with collecting multispectral satellite imagery, followed by preprocessing (104) to improve data quality through normalization and correction techniques. Relevant features are then extracted (106) from the processed data, which are used to design a Liquid Neural Network (108) architecture capable of capturing complex patterns. The network is subsequently implemented and trained using labeled data. Once trained, the model classifies the multispectral data (102) into different land categories (110) such as forest, water, urban, and agricultural land, and finally generates the classification output (112) in a structured form.
, Claims:CLAIMS
I/We Claim:
1. A method for designing and implementing a baseline Liquid Neural Network (108) architecture for multispectral Land Use and Land Cover (LULC) classification,
a. multispectral satellite data (102) comprising multiple spectral bands is collected for representing different land surface characteristics;
b. the collected multispectral satellite data (102) is preprocessed (104) by performing normalization and noise removal to improve data quality;
c. features are extracted (106) from the preprocessed (104) multispectral satellite data (102) to represent land characteristics;
d. a Liquid Neural Network (108) architecture is designed comprising adaptive neurons configured to process time-varying input data;
e. the Liquid Neural Network (108) architecture is implemented and trained using the extracted features to learn patterns associated with different land types;
f. the multispectral satellite data (102) is classified into one or more land categories (110) including forest, water, urban, and agricultural land; and
g. a classification output (112) is generated representing identified land use and land cover categories.

2. The method as claimed in claim 1, wherein the multispectral satellite data (102) comprises visible, near-infrared, and shortwave infrared spectral bands.

3. The method as claimed in claim 1, wherein the preprocessing (102) step comprises data normalization, noise filtering, and removal of inconsistencies in the multispectral data.
4. The method as claimed in claim 1, wherein the feature extraction (104) step comprises extraction of spatial and spectral features representing land characteristics.

5. The method as claimed in claim 1, wherein the Liquid Neural Network (106) architecture comprises adaptive neurons having time-varying parameters for processing dynamic input data.

6. The method as claimed in claim 1, wherein the training of the Liquid Neural Network (106) is performed using labeled multispectral datasets representing different land categories.

7. The method as claimed in claim 1, wherein the classification step includes land categorizing (110) into forest, water bodies, urban areas, and agricultural regions.

8. The method as claimed in claim 1, wherein the classification output (112) is presented in the form of maps, reports, or visual representations.

9. The method as claimed in claim 1, wherein the method is configured to handle variations in environmental conditions including seasonal changes, illumination differences, and noise in satellite imagery.

10. The method as claimed in claim 1, wherein the method provides a baseline implementation capable of optimization for improved accuracy and scalability.

 
6. DATE AND SIGNATURE
Dated this on 06th April 2026
Signature

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