AI, ML & IOT Hands-on in Agriculture

Goal: Ingest soil data and train a classifier to recommend crops.

• Phase 1: Data Ingestion & “Smart” Cleaning
  • Code: Import Pandas and load the Soil_Nutrients.csv.
  • Action: Handle missing values (imputation) and map text labels (e.g., “Rice”, “Maize”) to numbers using LabelEncoder.
  • Output: A clean, numeric feature matrix ready for AI.

• Phase 2: Training the Random Forest
  • Code: Import RandomForestClassifier from Scikit-Learn.
  • Action: Split data (80% Train / 20% Test). Write the training logic: model.fit().
  • Closure: Run a single prediction command on a specific soil sample to prove the model works.

Goal: Process leaf images and retrain a pre-existing AI model to detect disease.

• Phase 1: The Image Pipeline
  • Code: Import TensorFlow and ImageDataGenerator.
  • Action: Create a pipeline that automatically resizes and scales raw images from the dataset folder into a format the neural network accepts (224×224 pixels).

• Phase 2: Transfer Learning Implementation
  • Code: Load the pre-trained MobileNetV2.
  • Action: Freeze the “base” layers (so we don’t retrain the whole brain) and add a custom “Output Layer” for our specific disease classes.
  • Closure: Execute model.compile() and run 5 epochs of training. Watch the accuracy metric rise in real-time.

Goal: Format sensor logs and generate a future forecast line.

• Phase 1: Time-Series Formatting
  • Code: Import Prophet (or NeuralProphet).
  • Action: Convert standard sensor logs into the specific ds (Datestamp) and y (Value) format required by forecasting engines. Resample noisy data to hourly averages.

• Phase 2: Generating the Forecast
  • Code: Initialize and fit the Prophet model: m.fit(df).
  • Action: Create a “future dataframe” placeholder for the next 7 days.
  • Closure: Call model.predict(future) and plot the trend line to see the predicted moisture drop.

• Academic: Professors, Associate/Assistant Professors, and Guest Lecturers in Agriculture, Botany, or Environmental Sciences.
• Research: Ph.D. Scholars, Post-Doctoral Fellows, and M.Tech/M.Sc students specializing in Agronomy or Precision Farming.
• Industry: Agronomists, Soil Scientists, and Professionals working in Ag-Tech or Government Agricultural Departments.

• Coding: No prior programming experience is required. The workshop follows a “Zero-to-Code” approach.
• Mathematics: A basic understanding of statistical concepts (Mean, Median, Standard Deviation, Regression) is recommended.
• Data: Familiarity with handling data in Microsoft Excel or CSV formats.

• Hardware: A laptop with internet connectivity (Windows, Mac, or Linux).
• Account: A working Google (Gmail) account to access Google Colab.
• Browser: Latest version of Google Chrome or Firefox.

We’re here for you!

(+91) 120-4781-217