Datasets:

blackbeedrones
/

imav-2025-platform-dataset

license: apache-2.0
task_categories:
  - object-detection
tags:
  - drone
  - uav
  - imav
  - robotics
  - autonomous-landing
  - helipad-detection
  - nectar-sdk
  - IMAV-2025
size_categories:
  - 1K<n<10K
pretty_name: IMAV 2025 Platform Detection Dataset
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  features:
    - name: image
      dtype: image
    - name: image_id
      dtype: int64
    - name: width
      dtype: int32
    - name: height
      dtype: int32
    - name: objects
      struct:
        - name: id
          sequence: int64
        - name: bbox
          sequence:
            sequence: float32
            length: 4
        - name: category
          sequence:
            class_label:
              names:
                '0': platform
        - name: area
          sequence: float64

IMAV 2025 Platform Detection Dataset

Object detection dataset for landing platform detection in the IMAV 2025 Indoor Competition - Mission 4. The MAV must autonomously land on a moving 1m x 1m platform with an H-marking helipad, with optional smoke occlusion.

Trained model: blackbeedrones/imav-2025-platform

Competition Context

The 16th International Micro Air Vehicle Conference and Competition (IMAV 2025) took place in San Andres Cholula, Puebla, Mexico. The competition theme was "Search and Rescue", inspired by Mexico's seismic activity and the need for micro air vehicles in disaster response scenarios.

Mission 4: Land on Moving Platform with Smoke

The MAV must autonomously land on a moving platform:

Platform size: 1m x 1m
Lateral movement: up to 1m
Max speed: 0.5 m/s
Obstacle: Smoke machine (partial occlusion)

Target Object

Board: 1m x 1m square
Outer circle: 0.85m (black stroke)
Inner circle: 0.8m
H marking: 0.6m height, 0.35m width, 0.075m stroke

Dataset Structure

Split	Images
train	1043

Total images: 1043

Classes: platform

Annotation format: COCO (x_min, y_min, width, height)

Usage

Load with HuggingFace Datasets

from datasets import load_dataset

dataset = load_dataset("blackbeedrones/imav-2025-platform-dataset")
example = dataset["train"][0]
print(example["objects"])  # {'bbox': [...], 'category': [...]}

Visualize with bounding boxes

import torch
from torchvision.ops import box_convert
from torchvision.utils import draw_bounding_boxes
from torchvision.transforms.functional import pil_to_tensor, to_pil_image

example = dataset["train"][0]
categories = dataset["train"].features["objects"].feature["category"]

boxes_xywh = torch.tensor(example["objects"]["bbox"])
boxes_xyxy = box_convert(boxes_xywh, "xywh", "xyxy")
labels = [categories.int2str(x) for x in example["objects"]["category"]]

to_pil_image(
    draw_bounding_boxes(
        pil_to_tensor(example["image"]),
        boxes_xyxy,
        colors="red",
        labels=labels,
    )
)

Convert to COCO JSON (for training)

import json
from datasets import load_dataset

dataset = load_dataset("blackbeedrones/imav-2025-platform-dataset", split="train")
categories = dataset.features["objects"].feature["category"]

coco = {
    "images": [],
    "annotations": [],
    "categories": [
        {"id": i, "name": n} for i, n in enumerate(categories.names)
    ],
}

ann_id = 0
for row in dataset:
    coco["images"].append({
        "id": row["image_id"],
        "width": row["width"],
        "height": row["height"],
        "file_name": f"{row['image_id']}.jpg",
    })
    row["image"].save(f"images/{row['image_id']}.jpg")

    for bbox, cat, area in zip(
        row["objects"]["bbox"],
        row["objects"]["category"],
        row["objects"]["area"],
    ):
        coco["annotations"].append({
            "id": ann_id,
            "image_id": row["image_id"],
            "category_id": cat,
            "bbox": bbox,
            "area": area,
            "iscrowd": 0,
        })
        ann_id += 1

with open("annotations.json", "w") as f:
    json.dump(coco, f)

Train with Nectar SDK

from nectar.ai.detection import Detector, TrainingConfig

detector = Detector("yolo11n.pt")
detector.load()
result = detector.train(TrainingConfig(
    dataset_path="path/to/converted/dataset",
    epochs=100,
    push_to_hub=True,
    hub_model_id="blackbeedrones/imav-2025-platform",
))

Train with Ultralytics (YOLO format)

from datasets import load_dataset
from pathlib import Path

dataset = load_dataset("blackbeedrones/imav-2025-platform-dataset")
categories = dataset["train"].features["objects"].feature["category"]

for split_name, split_data in dataset.items():
    img_dir = Path(f"yolo_dataset/images/{split_name}")
    lbl_dir = Path(f"yolo_dataset/labels/{split_name}")
    img_dir.mkdir(parents=True, exist_ok=True)
    lbl_dir.mkdir(parents=True, exist_ok=True)

    for row in split_data:
        fname = f"{row['image_id']}"
        row["image"].save(img_dir / f"{fname}.jpg")
        w, h = row["width"], row["height"]

        with open(lbl_dir / f"{fname}.txt", "w") as f:
            for bbox, cat in zip(row["objects"]["bbox"], row["objects"]["category"]):
                x_center = (bbox[0] + bbox[2] / 2) / w
                y_center = (bbox[1] + bbox[3] / 2) / h
                bw, bh = bbox[2] / w, bbox[3] / h
                f.write(f"{cat} {x_center} {y_center} {bw} {bh}\n")