Evaluation Data (Image Retrieval)
SOP
#Evaluation Metrics - R@1, R@10, R@100, R@1000
Data Description:
The SOP dataset contains 120,053 product images with 22,634 categories. Among these categories, 5,659 categories (29,722 images) are used for training, 5,659 categories (29,829 images) are used for validation, and the remaining 11,316 categories (60,502 images) are used for testing.
Dataset structure:
Amount of source data:
The dataset is divided into auxiliary training set (29,722), validation set (29,829), and testing set (60,502).
Amount of test data:
All 60,502 test examples from the source dataset test set.
Data detail:
| KEYS | EXPLAIN |
|---|---|
| images | RGB images used as input |
| labels | Category label of the image |
Sample of source dataset:
RGB image：
label：
bicycle
Paper Citation:
@inproceedings{oh2016deep,
title={Deep metric learning via lifted structured feature embedding},
author={Oh Song, Hyun and Xiang, Yu and Jegelka, Stefanie and Savarese, Silvio},
booktitle={Computer Vision and Pattern Recognition},
pages={4004--4012},
year={2016}
}Dataset Copyright Usage Information:
iNaturalist-2018
#Evaluation Metrics - R@1, R@4, R@16, R@32
Data Description
The iNaturalist-2018 dataset has 8,142 species categories, with 437,513 training images, 24,426 validation images, and 149,394 test images. Andrew Brown performed sampling and split the dataset into 5,690 categories (325,846 images) for training and 2,452 categories (136,093 images) for testing. Based on the partitioning by Andrew Brown, in the original training set, 2,848 categories are used for training, and 2,842 categories are used for validation. The original testing set remains unchanged.
Dataset structure:
Amount of source data:
The dataset is divided into auxiliary training set (165,229), validation set (160,617), and testing set (136,093).
Amount of test data:
All 136,093 test examples from the test set split by Andrew Brown.
Data detail:
| KEYS | EXPLAIN |
|---|---|
| images | RGB images used as input |
| labels | Category label of the image |
Sample of source dataset:
RGB image：
label：
2229
Paper Citation:
@inproceedings{van2018inaturalist,
title={The inaturalist species classification and detection dataset},
author={Van Horn, Grant and Mac Aodha, Oisin and Song, Yang and Cui, Yin and Sun, Chen and Shepard, Alex and Adam, Hartwig and Perona, Pietro and Belongie, Serge},
booktitle={Computer Vision and Pattern Recognition},
pages={8769--8778},
year={2018}
}