Skip to content

Dataset · LouisChen15

Annotated construction sites

LouisChen15/ConstructionSite

10,013 construction site images with annotations for evaluating scene understanding in vision-language models.

What's inside

The dataset contains 10,013 construction site images and their annotations, with 7,009 images in the train split and 3,004 in the test split. Written in English, the dataset is designed to evaluate how well vision-language models understand construction site imagery.

What you can do with it

  • Evaluating VLMs on construction scene understanding
  • Fine-tuning image-to-text models on construction sites
  • Extracting image features for construction monitoring
Tasks
Image → text, Image feature extraction
Languages
English
Size
10K–100K rows
Downloads this month
1.2K
Likes
84
Access
on request on Hugging Face

How to load it

from datasets import load_dataset

ds = load_dataset("LouisChen15/ConstructionSite")