blackbeedrones/imav-2025-platform

IMAV 2025 Platform Detection - YOLOv11n

Platform detection model for IMAV 2025 Indoor Competition - Mission 4.

Competition Context

The 16th International Micro Air Vehicle Conference and Competition (IMAV 2025) took place in San Andrés 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.

Target Object

Platform Specifications:

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

Mission 4: Land on Moving Platform with Smoke

The MAV must autonomously land on a moving platform:

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

Scoring:

Task	Points
No landing	0
Landing (stationary)	2
Landing (moving platform)	+3
Landing (with smoke)	+3

Performance

Metric	Value
mAP@50	0.995
mAP@50-95	0.973
Precision	0.996
Recall	0.989

Training Curves

Confusion Matrix

Validation Predictions

Model Formats

Format	File	Use Case
PyTorch	platform_yolov11n.pt	Training, fine-tuning
ONNX	platform_yolov11n.onnx	Cross-platform inference
TensorRT	platform_yolov11n.engine	Jetson Orin Nano Super

Training Configuration

Parameter	Value
Base model	yolo11n.pt
Epochs	100
Image size	640×640
Batch	Auto
Optimizer	Auto
LR	0.01 → 0.01 (cosine)
Augmentation	Mosaic, RandAugment
Dropout	0.05

Full config: train/args.yaml

Usage

Nectar SDK

from nectar.ai.detection import Detector

detector = Detector("blackbeedrones/imav-2025-platform:best.pt")
detector.load()

result = detector.detect(image, conf=0.5)
for det in result:
    print(f"Platform: {det.confidence:.2f} at {det.center}")

Ultralytics

from ultralytics import YOLO

model = YOLO("best.pt")
results = model.predict(image, conf=0.5)

References

• IMAV 2025
• Rulebook
• Nectar SDK
• Dataset