🛰️ Antennas-TDT: Telecom Tower Detection and Classification Pipeline

Antennas-TDT is a Python-based tool for the automatic detection, segmentation, and classification of antennas on telecommunication towers.
It integrates a NeRF-based 3D reconstruction pipeline powered by Nerfstudio with a custom geospatial and deep learning toolkit (TDT-Tool).

Developed by the TIDOP Research Group — University of Salamanca (Spain).

---

📖 Overview

The Antennas-TDT workflow consists of two components:

Component	Description
nerfstudio_tdt	A Conda environment with Nerfstudio, PyTorch, and CUDA 11.8. Used to process COLMAP and NeRF training.
tdt-tool	The main Python package implementing antenna detection, segmentation, and classification. Runs on top of the nerfstudio_tdt environment.

The pipeline automatically performs:

1. 3D reconstruction (NeRF)
2. Georeferencing and cropping
3. Antenna detection (YOLO-based)
4. Segmentation and classification
5. Final 3D model export in PLY format

---

🧩 Installation

Requirements

• Windows 10/11 (64-bit)
• NVIDIA GPU compatible with CUDA ≥ 11.8
• Anaconda or Miniconda
• Latest NVIDIA drivers (nvidia-smi should work)
• CloudCompare ≥ 2.12 — required for point cloud post-processing and visualization

⚠️ CloudCompare is an external tool, not a Python package.
Download and install it manually from the official website.
It is recommended to add its installation folder to your system PATH so it can be called from the command line.

---

🅰️ Step 1 — Install the Nerfstudio GPU environment

1. Open an Anaconda Prompt.
2. Navigate to the project folder:

cd Antennas-TDT

3. Create the environment using the BAT file:

install_nerfstudio_gpu.bat

4. Activate it and verify CUDA and PyTorch (optional):

conda activate nerfstudio_tdt
python -c "import torch; print('PyTorch', torch.__version__, 'CUDA', torch.version.cuda, 'GPU available?', torch.cuda.is_available())"

✅ Expected output:

PyTorch 2.1.2 CUDA 11.8 GPU available? True

If the verification fails, check that your NVIDIA driver is up to date and supports CUDA 11.8.

🅱️ Step 2 — Install TDT-Tool

1. Open an Anaconda Prompt.
2. Navigate to the project folder:

cd Antennas-TDT

3. Create the environment using the YML file:

conda env create -f env-main_tdt.yml

4. Activate it:

conda activate tdt-tool

🚀 Usage Once installed, you can run the full pipeline:

conda run -n tdt-tool python main.py "C:\Path\To\Your\ProjectFolder"

Expected folder structure:

ProjectFolder/
├── images/
└── (output generated automatically)