Datasets:

Modalities:
Image
Formats:
parquet
Languages:
English
Size:
< 1K
Libraries:
Datasets
pandas
License:
pucks / README.md
jagennath-hari's picture
update README.md
a1c50d8
metadata
language:
  - en
pretty_name: Pucks surrogate Laserweeder
license: other
license_name: laser-dataset-replication-license-v1.0
license_link: LICENSE
tags:
  - AgTech
annotations_creators:
  - machine-generated
language_creators:
  - found
size_categories:
  - n<1K
source_datasets:
  - original
task_categories:
  - object-detection
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*
dataset_info:
  features:
    - name: image_id
      dtype: int32
    - name: image
      dtype: image
    - name: annotated_image
      dtype: image
    - name: resolution
      sequence: int32
      length: 2
    - name: bboxes
      sequence:
        sequence: float32
        length: 4
    - name: category_ids
      sequence: int32
  splits:
    - name: train
      num_bytes: 316722721
      num_examples: 118
    - name: validation
      num_bytes: 48332168
      num_examples: 18
    - name: test
      num_bytes: 24279368
      num_examples: 9
  download_size: 389352498
  dataset_size: 389334257

Pucks dataset

Dataset Overview

The Pucks Dataset is part of the L&Aser surrogate target recognition pipeline developed by Laudando & Associates LLC. This dataset contains high-resolution field images annotated with bounding boxes around small circular surrogate targets (“pucks”) used to simulate real weed targets for the L&Aser weeding system.

Contents

  • Raw RGB images collected in field conditions using a NVIDIA Jetson and monocular camera

  • Bounding boxes in COCO-style format generated using a neural network

  • Annotated visual overlays for quick inspection

Preview Format

This dataset is optimized for visual inspection on the Hugging Face Hub, with an additional annotated_image column for rapid validation. The raw COCO JSON and images are included in their original form for reproducibility and training.

Example Use

from datasets import load_dataset

ds = load_dataset("Laudando-Associates-LLC/pucks", split="train")
img = ds[0]["image"]
ann = ds[0]["bboxes"]

Recording Date

The data was originally collected in ROS 2 bag format in April 2024 during controlled field trials conducted by Laudando & Associates LLC. This dataset is a stripped-down version containing only the extracted RGB images and their corresponding object detection annotations.

License

This dataset is provided under the L&Aser Dataset Replication License (Version 1.0). See LICENSE for full terms.