Detecting and counting trees from satellite images

This project implements density map based method for dense object counting, specifically applied to tree counting in aerial images.

Features

• Utilizes a U-Net architecture for density map prediction
• Implements Density map generation (DMG) model for ground truth generation
• Supports training, validation, and testing phases
• Includes visualization tools for density maps

Installation & Setup

1. Run the setup script:

   sh ./startup.sh

   This script will:

   • Create and activate a virtual environment
   • Install all required packages
   • Download the Yosemite Tree Dataset
   • Preprocess the dataset

Pretrained model

The pretrained models are included in the ./phases directory.

Training

To train the model, run:

python train.py

You can override config parameters using command-line arguments:

python train.py  --override data_dir=./new_data_path lr=1e-4

Testing

To test the model:

python test.py --model_folder ./path_to_model_folder

Configuration

The config.json file contains all the hyperparameters and settings for the model. You can modify this file to change the model's behavior.

Project Structure

• models/: Contains the U-Net and DMG model implementations
• datasets/: Includes the TreeCountingDataset class
• utils/: Helper functions and classes for training and evaluation
• train.py: Main script for training the model
• test.py: Script for evaluating the model on the test set
• visualizer.py: Script for visualizing density maps