Back to datasets
Dataset assetOpen Source CommunityImage ProcessingSalient Object Detection
PASCAL VOC 2010
The dataset contains images and their corresponding saliency maps for training and validating saliency detection models based on recurrent U‑Net. It is packaged into two .npy files, one for images and one for the associated saliency maps.
Source
github
Created
Dec 24, 2021
Updated
Dec 24, 2021
Signals
176 views
Availability
Linked source ready
Overview
Dataset description and usage context
Dataset Overview
Dataset Name
PASCAL VOC 2010
Dataset Content
- X_UNET_recurr_prior_map.npy: Contains all image data with shape (num_images, 224, 224, 4). The first three channels store RGB images; the fourth channel holds the saliency prior for each image.
- y_UNET_recurr_prior_map.npy: Contains ground‑truth image data with shape (num_images, T, 224, 224, 1). Here, T denotes the recurrent U‑Net time steps; for this dataset, T = 3.
Dataset Storage
- X_UNET_recurr_prior_map.npy: Google Drive link
- y_UNET_recurr_prior_map.npy: Google Drive link
Dataset Usage
The dataset is intended for training a recurrent U‑Net model for saliency detection. It has been pre‑processed and compressed; users should ensure correct file paths in their notebooks.
Need downstream help?
Pair the dataset with AI analysis and content workflows.
Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.