Research map / Detection and segmentation / Weakly supervised detection and segmentation
ELDET: Early-Learning Distillation with Noisy Labels for Object Detection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes ELDET, an early-learning distillation method for training object detectors on datasets with both categorization and localization label noise.
Problem
Object detectors trained on large-scale annotated datasets suffer from two co-occurring label noises: incorrect class labels (categorization noise) and inaccurate bounding boxes (localization noise).
Approach
Builds on early-learning behavior of networks under noisy labels, distilling knowledge to handle both categorization and localization noise jointly in object detection. Only the abstract is available, so the specific mechanism is not described here.
Closest papers by meaning
- Distilling Image Classifiers in Object Detectors (NeurIPS 2021 · similarity 0.89)
- Scale-Equivalent Distillation for Semi-Supervised Object Detection (CVPR 2022 · similarity 0.89)
- Object Detection Benchmarks are Incomplete: The Role of Label Errors and Annotation Uncertainty (NeurIPS 2026 · similarity 0.87)
- LEOD: Label-Efficient Object Detection for Event Cameras (CVPR 2024 · similarity 0.87)
- SimLTD: Simple Supervised and Semi-Supervised Long-Tailed Object Detection (CVPR 2025 · similarity 0.87)
- DetKDS: Knowledge Distillation Search for Object Detectors (ICML 2024 · similarity 0.87)
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.