Datasetpublic

RedBall/seoul-urban-park-analysis-by-llm

This dataset contains analysis results of 29,866 urban park images. The analysis is done by GPT-4. The elements for analysis are as follows: "elements": All physical elements visible in the photo, and the proportion of each element in the photo (summing up to 1). For each type of element (e.g., structures, ground, objects), provide a descriptive analysis, stating them by specific pronouns. For example, you should not state a building just as a building, but more precisely like a pavilion or… See the full description on the dataset page: https://huggingface.co/datasets/RedBall/seoul-urban-park-analysis-by-llm.

sourceHugging Facemitupdated 2y agoView on Hugging Face

1likes1downloads

Dataset Card

This dataset contains analysis results of 29,866 urban park images. The analysis is done by GPT-4.

The elements for analysis are as follows:

—"elements": All physical elements visible in the photo, and the proportion of each element in the photo (summing up to 1). For each type of element (e.g., structures, ground, objects), provide a descriptive analysis, stating them by specific pronouns. For example, you should not state a building just as a building, but more precisely like a pavilion or canopy.
—"isuserexist": Whether people exist in the photo.
—"user_activities": If there are people, what activities they are doing.
—"user_emotions": If there are people, the emotions of the people.
—"user_group": If there are people, in what group they are.
—"user_type": If there are people, the gender and age group.
—"isanimalexist": Whether animals appear.
—"animals": If there are animals, what kinds of animals there are.
—"time": The time of day the photo was taken.
—"season": The season the photo was taken. Choose one of 4 seasons.
—"image_emotions": The emotions felt from the photo.

As a side note, this dataset does not contain images because of privacy issue. There are only analysis results.