Dataset and approach for helmet-wearing detection with edge devices at construction sites

Abstract

{{ abstract }}

Examples

examples-picture

Video for Previewing Results

Citation

{{ bibCite }}

Download Dataset

The benchmark dataset HWD2024 is available in the COCO (Common Objects in Context) format, which is used by the most object detection datasets.
Please contact Liao Longlong via the Email liaolonglong@fzu.edu.cn if you have any suggestion or question about improving and downloading this benchmark dataset.
The annotation files of samples in its test set are excluded from the download dataset since you are required to evaluate your models by uploading test results and evaluating with our online system. Online model evaluation makes the results are more fair to compare with each other. Once the paper is accepted, the dataset will be available to download.

Model Evaluation

    To evaluate the model accuracy (AP per class and mAP) based on its JSON-formatted prediction results:
  1. Click the 'Upload JSON file' button to upload a JSON file that contains the prediction results of a model. The demo JSON file is available at here, to illustrate the data format of predict results.
  2. Email verification is required to confirm your identity; it must be an education Email.
  3. Click on the "Evaluate & Show Results" button to obtain the evaluation results. If your model results are outstanding, you can choose whether to add the results to the following table.
{{ uploadedFiles[0].name }}

Model Evaluation Results

This following table displays the evaluation results of state-of-the-art models:

Model Metric Overall blue white yellow red none correct person orange wrong
Loading...