# Soft Sampling for Robust Object Detection

arXiv: https://arxiv.org/abs/1806.06986 · HTML: https://mlsanity.com/paper/1806.06986/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Jun 2018.

## Built on by (most cited first, 22 of 22)

- [Object Detection in Aerial Images: A Large-Scale Benchmark and Challenges](https://mlsanity.com/paper/2102.12219/) (2021)
- [Data-Uncertainty Guided Multi-Phase Learning for Semi-Supervised Object Detection](https://mlsanity.com/paper/2103.16368/) (2021)
- [A Step Toward More Inclusive People Annotations for Fairness](https://mlsanity.com/paper/2105.02317/) (2021)
- [Identifying Label Errors in Object Detection Datasets by Loss Inspection](https://mlsanity.com/paper/2303.06999/) (2023)
- [SparseDet: Improving Sparsely Annotated Object Detection with Pseudo-positive Mining](https://mlsanity.com/paper/2201.04620/) (2022)
- [Box-Level Active Detection](https://mlsanity.com/paper/2303.13089/) (2023)
- [Calibrated Teacher for Sparsely Annotated Object Detection](https://mlsanity.com/paper/2303.07582/) (2023)
- [Active learning using weakly supervised signals for quality inspection](https://mlsanity.com/paper/2104.02973/) (2021)
- [Semi-Supervised and Long-Tailed Object Detection with CascadeMatch](https://mlsanity.com/paper/2305.14813/) (2023)
- [Are Dense Labels Always Necessary for 3D Object Detection from Point Cloud?](https://mlsanity.com/paper/2403.02818/) (2024)
- [Dynamic Supervisor for Cross-dataset Object Detection](https://mlsanity.com/paper/2204.00183/) (2022)
- [Soft-labeling Strategies for Rapid Sub-Typing](https://mlsanity.com/paper/2209.12684/) (2022)
- [Robust Object Detection in Remote Sensing Imagery with Noisy and Sparse Geo-Annotations (Full Version)](https://mlsanity.com/paper/2210.12989/) (2022)
- [An End-to-End Framework For Universal Lesion Detection With Missing Annotations](https://mlsanity.com/paper/2303.15024/) (2023)
- [Tackling the Incomplete Annotation Issue in Universal Lesion Detection Task By Exploratory Training](https://mlsanity.com/paper/2309.13306/) (2023)
- [Leveraging Sparse Annotations for Leukemia Diagnosis on the Large Leukemia Dataset](https://mlsanity.com/paper/2504.02602/) (2025)
- [Decoupling Classifier for Boosting Few-shot Object Detection and Instance Segmentation](https://mlsanity.com/paper/2505.14239/) (2025)
- [TagMe: GPS-Assisted Automatic Object Annotation in Videos](https://mlsanity.com/paper/2103.13428/) (2021)
- [ComplETR: Reducing the cost of annotations for object detection in dense scenes with vision transformers](https://mlsanity.com/paper/2209.05654/) (2022)
- [Soft Labels for Rapid Satellite Object Detection](https://mlsanity.com/paper/2212.00585/) (2022)
- [Towards Sparsely Annotated Open-World Object Detection](https://mlsanity.com/paper/2608.12714/) (2026)
- [Background-Free Objectness Learning for Class-Agnostic Detection](https://mlsanity.com/paper/2608.29232/) (2026)

## Abstract (arXiv)

We study the robustness of object detection under the presence of missing annotations. In this setting, the unlabeled object instances will be treated as background, which will generate an incorrect training signal for the detector. Interestingly, we observe that after dropping 30% of the annotations (and labeling them as background), the performance of CNN-based object detectors like Faster-RCNN only drops by 5% on the PASCAL VOC dataset. We provide a detailed explanation for this result. To further bridge the performance gap, we propose a simple yet effective solution, called Soft Sampling. Soft Sampling re-weights the gradients of RoIs as a function of overlap with positive instances. This ensures that the uncertain background regions are given a smaller weight compared to the hardnegatives. Extensive experiments on curated PASCAL VOC datasets demonstrate the effectiveness of the proposed Soft Sampling method at different annotation drop rates. Finally, we show that on OpenImagesV3, which is a real-world dataset with missing annotations, Soft Sampling outperforms standard detection baselines by over 3%.