Saved in:
Bibliographic Details
Main Authors: Becker, Stefan, Bayer, Jens, Hug, Ronny, Hübner, Wolfgang, Arens, Michael
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2311.09768
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910437528829952
author Becker, Stefan
Bayer, Jens
Hug, Ronny
Hübner, Wolfgang
Arens, Michael
author_facet Becker, Stefan
Bayer, Jens
Hug, Ronny
Hübner, Wolfgang
Arens, Michael
contents Data pooling offers various advantages, such as increasing the sample size, improving generalization, reducing sampling bias, and addressing data sparsity and quality, but it is not straightforward and may even be counterproductive. Assessing the effectiveness of pooling datasets in a principled manner is challenging due to the difficulty in estimating the overall information content of individual datasets. Towards this end, we propose incorporating a data source prediction module into standard object detection pipelines. The module runs with minimal overhead during inference time, providing additional information about the data source assigned to individual detections. We show the benefits of the so-called dataset affinity score by automatically selecting samples from a heterogeneous pool of vehicle datasets. The results show that object detectors can be trained on a significantly sparser set of training samples without losing detection accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2311_09768
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Utilizing dataset affinity prediction in object detection to assess training data
Becker, Stefan
Bayer, Jens
Hug, Ronny
Hübner, Wolfgang
Arens, Michael
Computer Vision and Pattern Recognition
Data pooling offers various advantages, such as increasing the sample size, improving generalization, reducing sampling bias, and addressing data sparsity and quality, but it is not straightforward and may even be counterproductive. Assessing the effectiveness of pooling datasets in a principled manner is challenging due to the difficulty in estimating the overall information content of individual datasets. Towards this end, we propose incorporating a data source prediction module into standard object detection pipelines. The module runs with minimal overhead during inference time, providing additional information about the data source assigned to individual detections. We show the benefits of the so-called dataset affinity score by automatically selecting samples from a heterogeneous pool of vehicle datasets. The results show that object detectors can be trained on a significantly sparser set of training samples without losing detection accuracy.
title Utilizing dataset affinity prediction in object detection to assess training data
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2311.09768