ReImageNet paper finds 12% errors in original ImageNet labels
ducha_aiki · x · 2026-08-21
A paper titled 'Doomed to Re-Annotate, Forever: The ImageNet Story' introduces ReImageNet, a comprehensive re-annotation effort for the ImageNet-1k validation set. The study reveals that approximately 12% of the original labels are incorrect, 33.3% of images are multi-label, and 3.8% contain no object from an ImageNet-1k class. With the new labels, Top-1 accuracy increases by up to 1.2% for supervised models and 5-6% for MLLMs. The authors argue that annotation at this scale cannot be completed in one pass and requires iterative refinement through human-LLM collaboration.
Related event: ReImageNet Released: Re-Annotation Exposes 12% Label Errors in ImageNet(3 posts)→
More from Research
- Transformer trained on flying bat recordings yields hippocampus 'digital twin' revealing hidden structure — patrickmineault · 2026-10-16
- One year of Recursive Language Models: GPT-5-mini beats GPT-5 by 110% on 132k-token tasks, cheaper too — a1zhang · 2026-10-16
- MIT PhD Sharut Gupta joins DeepMind; Isola hails her representational alignment defense talk — phillip_isola · 2026-10-16
- 36 primitives, just 10 parameters each: a remarkably precise T-pose body fit — letandrewcook · 2026-10-16
- Interp researcher pushes back: if you can't simulate a neuron, you haven't explained it — nabla_theta · 2026-10-16
- MLP-Splatting (ECCV 2026) code released: MLPs as light-field primitives instead of Gaussians — AjdDavison · 2026-10-16