Are Labels Needed for Incremental Instance Learning?

CVPRW 2023 pp. 2401-2409

doi:10.1109/CVPRW59228.2023.00238 /cvprw/2023/kilickaya2023cvprw-labels/

Abstract

In this paper, we learn to classify visual object instances, incrementally and via self-supervision (self-incremental). Our learner observes a single instance at a time, which is then discarded from the dataset. Incremental instance learning is challenging, since longer learning sessions exacerbate forgetfulness, and labeling instances is cumbersome. We overcome these challenges via three contributions: i). We propose VINIL, a self-incremental learner that can learn object instances sequentially, ii). We equip VINIL with self-supervision to by-pass the need for instance la-belling, iii). We compare VINIL to label-supervised variants on two large-scale benchmarks [6], [32], and show that VINIL significantly improves accuracy while reducing forgetfulness.

PDF CVPRW Semantic Scholar

Cite

Text

Kilickaya and Vanschoren. "Are Labels Needed for Incremental Instance Learning?." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2023. doi:10.1109/CVPRW59228.2023.00238

Markdown

[Kilickaya and Vanschoren. "Are Labels Needed for Incremental Instance Learning?." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2023.](https://mlanthology.org/cvprw/2023/kilickaya2023cvprw-labels/) doi:10.1109/CVPRW59228.2023.00238

BibTeX

@inproceedings{kilickaya2023cvprw-labels,
  title     = {{Are Labels Needed for Incremental Instance Learning?}},
  author    = {Kilickaya, Mert and Vanschoren, Joaquin},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},
  year      = {2023},
  pages     = {2401-2409},
  doi       = {10.1109/CVPRW59228.2023.00238},
  url       = {https://mlanthology.org/cvprw/2023/kilickaya2023cvprw-labels/}
}