CoolFace
Datasetpublic

lll-a-p/fpv-drone-detection-dataset

FPV Analog Video Signal Dataset for Real-time Detection Description This dataset contains captured analog CVBS (Color, Video, Blanking, and Sync) signals from FPV drones, specifically NTSC standard. The data was collected to support the development of signal processing algorithms for resource-constrained embedded systems (ESP32 and STM32). Dataset Structure The dataset is provided as a master index file and several compressed archives containing raw… See the full description on the dataset page: https://huggingface.co/datasets/lll-a-p/fpv-drone-detection-dataset.

sourceHugging Facemitupdated 5mo agoView on Hugging Face
0likes82downloads
Dataset Card

FPV Analog Video Signal Dataset for Real-time Detection

Description

This dataset contains captured analog CVBS (Color, Video, Blanking, and Sync) signals from FPV drones, specifically NTSC standard. The data was collected to support the development of signal processing algorithms for resource-constrained embedded systems (ESP32 and STM32).

Dataset Structure

The dataset is provided as a master index file and several compressed archives containing raw CSV signal data.

1. dataset_index.csv

The metadata file containing the following columns:

  • —idx: Sample identifier.
  • —freq_ghz: The carrier frequency band (1.2, 3.3, or 5.8 GHz).
  • —category: Classification label (clos, degraded, non_reconstructible, noise_floor, interference).
  • —img_path: Path to the visual spectrogram/time-domain plot (found in imgs.zip).
  • —csv_vtx_path: Path to the source signal (Transmitter side).
  • —csv_vrx_path: Path to the captured signal (Receiver side).

2. Data Archives

  • —`12GHz.zip`: Captures from the 1.2 GHz frequency band.
  • —`33GHz.zip`: Captures from the 3.3 GHz frequency band.
  • —`58GHz.zip`: Captures from the 5.8 GHz frequency band.
  • —`imgs.zip`: Supporting JPG visualizations for each sample.

Acquisition Setup

  • —Hardware: Custom RF front-end connected to Osciloscope.
  • —Sampling Rates: 5MSps
  • —Signal Conditions: Range from Clear Line of Sight (CLOS) to signals buried in the noise floor ($SNR < 0 \text{ dB}$).

Usage

To use the dataset with the provided paths in the CSV index, extract the archives into a data/ directory:

bash
mkdir data
unzip 12GHz.zip -d ./data/12GHz/
unzip 33GHz.zip -d ./data/33GHz/
unzip 58GHz.zip -d ./data/58GHz/
unzip imgs.zip -d ./data/imgs/