Dataset · hanchong
Super-resolved infrared maritime vessel images
hanchong/real-infrared-maritime-vessel-dataset-super-resolved
Tens of thousands of PNG infrared ship images enhanced with HYPIR for classification.
What's inside
The dataset includes cropped real infrared images of maritime vessels (PNG format) enhanced via super-resolution using HYPIR. Data is provided in two variants: restored with and without a class-name text prompt. The dataset is split into train, val, and test sets across 7 classes (0 to 6), with the training set alone containing 25,045 images, and annotations provided in CSV and TXT files.
What you can do with it
- Training computer vision models for infrared maritime vessel recognition and classification
- Comparing the impact of text prompts (prompt-guided restoration) on post-upscale classifier performance
- Tasks
- Image classification
- Size
- 10K–100K rows
- Downloads this month
- 29.5K
- Likes
- 0
How to load it
from datasets import load_dataset
ds = load_dataset("hanchong/real-infrared-maritime-vessel-dataset-super-resolved")