This model is a YOLO11 object detector trained for a real Operational Technology (OT) vision use case: detecting forklifts and people in industrial/warehouse imagery to support pedestrian/forklift collision-risk monitoring.

Application

Industrial sites, warehouses, and plants use cameras for safety monitoring. A detector for forklift and person is a core perception component for zone intrusion alerts, near-miss analytics, and driver/pedestrian warning systems.

Training data

Dataset: keremberke/forklift-object-detection
License: CC BY 4.0
Classes: ['forklift', 'person']
Audit: {"train": {"images": 2, "boxes": 4, "classes": {"forklift": 2, "person": 2}, "boxes_per_image_min": 2, "boxes_per_image_max": 2, "boxes_per_image_mean": 2.0, "bad_boxes_skipped": 0}, "val": {"images": 2, "boxes": 5, "classes": {"forklift": 4, "person": 1}, "boxes_per_image_min": 2, "boxes_per_image_max": 3, "boxes_per_image_mean": 2.5, "bad_boxes_skipped": 0}, "test": {"images": 2, "boxes": 2, "classes": {"forklift": 2}, "boxes_per_image_min": 1, "boxes_per_image_max": 1, "boxes_per_image_mean": 1.0, "bad_boxes_skipped": 0}}

The original COCO annotations were converted to Ultralytics YOLO format. Bounding boxes were clipped to image bounds when necessary; invalid zero-area boxes were skipped.

Training recipe

Base model: yolo11n.pt from Ultralytics/YOLO11
Epochs: 1
Image size: 640
Batch size: 2
Optimizer: SGD
LR0: 0.01
Momentum: 0.937
Weight decay: 0.0005
Patience: 30

Validation metrics

json

{
  "metrics/precision(B)": 0.007017543859649123,
  "metrics/recall(B)": 1.0,
  "metrics/mAP50(B)": 0.2238235294117648,
  "metrics/mAP50-95(B)": 0.08390324665431029,
  "fitness": 0.08390324665431029
}

Usage

python

from ultralytics import YOLO

model = YOLO("best.pt")
results = model("factory_or_warehouse_image.jpg")

Monitoring

Training was logged with Trackio project ot-yolo-forklift-safety at Space: https://huggingface.co/spaces/akdve/ml-intern-ot-yolo-smoke

Generated by ML Intern

This model repository was generated by ML Intern, an agent for machine learning research and development on the Hugging Face Hub.

Try ML Intern: https://smolagents-ml-intern.hf.space
Source code: https://github.com/huggingface/ml-intern

akdve/ot-forklift-person-yolo11s-smoke

OT Forklift-Person YOLO11 Detector