Research map / Detection and segmentation / Weakly supervised detection and segmentation
DatasetGAN: Efficient Labeled Data Factory With Minimal Human Effort
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train on DatasetGAN-generated labeled images instead of the semi-supervised baseline data | DatasetGAN · ADE-Car-12 | mIoU: 34.82 -> 45.64 | Helped |
| train on DatasetGAN-generated labeled images instead of the semi-supervised baseline data | DatasetGAN · ADE-Car-5 | mIoU: 48.76 -> 57.77 | Helped |
| train on DatasetGAN-generated labeled images instead of the semi-supervised baseline data | DatasetGAN · Car-20 | mIoU: 44.51 $\pm$ 0.94 -> 62.33 $\pm$ 0.55 | Helped |
| train on DatasetGAN-generated labeled images instead of the semi-supervised baseline data | DatasetGAN · CelebA-Mask-8 | mIoU: 65.53 -> 70.01 | Helped |
| train on DatasetGAN-generated labeled images instead of the semi-supervised baseline data | DatasetGAN · Face-34 | mIoU: 48.17 $\pm$ 0.66 -> 53.46 $\pm$ 1.21 | Helped |
| train on DatasetGAN-generated labeled images instead of the semi-supervised baseline data | DatasetGAN · Bird-11 | mIoU: 25.04 $\pm$ 0.29 -> 36.76 $\pm$ 2.11 | Helped |
| train on DatasetGAN-generated labeled images instead of the semi-supervised baseline data | DatasetGAN · Cat-16 | mIoU: 24.85 $\pm$ 0.35 -> 31.26 $\pm$ 0.71 | Helped |
| train on DatasetGAN-generated labeled images instead of the semi-supervised baseline data | DatasetGAN · Bedroom-19 | mIoU: 30.15 $\pm$ 0.52 -> 36.83 $\pm$ 0.54 | Helped |
| train keypoint model on DatasetGAN-generated annotations instead of transfer learning | DatasetGAN · Car-20 | PCK th-15: 43.54 -> 79.91 | Helped |
| train keypoint model on DatasetGAN-generated annotations instead of transfer learning | DatasetGAN · Car-20 | L2 Loss: 4.4 $\times 10^{-4}$ -> 2.4 $\times 10^{-4}$ | Helped |
| increase synthesized training set from 3K to 10K images (ablation) | DatasetGAN · ADE-Car-12 | mIoU: 43.34 -> 44.60 | Helped |
| increase synthesized training set from 10K to 20K images (ablation) | DatasetGAN · ADE-Car-12 | mIoU: 44.60 -> 45.04 | Helped |
Closest papers by meaning
- BigDatasetGAN: Synthesizing ImageNet With Pixel-Wise Annotations (CVPR 2022 · similarity 0.92)
- Repurposing GANs for One-Shot Semantic Part Segmentation (CVPR 2021 · similarity 0.90)
- Semantic Segmentation With Generative Models: Semi-Supervised Learning and Strong Out-of-Domain Generalization (CVPR 2021 · similarity 0.90)
- GANSeg: Learning To Segment by Unsupervised Hierarchical Image Generation (CVPR 2022 · similarity 0.88)
- Learning to Annotate Part Segmentation with Gradient Matching (ICLR 2022 · similarity 0.88)
- EditGAN: High-Precision Semantic Image Editing (NeurIPS 2021 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.