Sampling Techniques for Large-Scale Object Detection From Sparsely Annotated Objects

@article{Niitani2019SamplingTF,
  title={Sampling Techniques for Large-Scale Object Detection From Sparsely Annotated Objects},
  author={Yusuke Niitani and Takuya Akiba and Tommi Kerola and Toru Ogawa and Shotaro Sano and Shuji Suzuki},
  journal={2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2019},
  pages={6503-6511}
}
Efficient and reliable methods for training of object detectors are in higher demand than ever, and more and more data relevant to the field is becoming available. However, large datasets like Open Images Dataset v4 (OID) are sparsely annotated, and some measure must be taken in order to ensure the training of a reliable detector. In order to take the incompleteness of these datasets into account, one possibility is to use pretrained models to detect the presence of the unverified objects… 

Figures and Tables from this paper

Sparsely Annotated Object Detection: A Region-based Semi-supervised Approach
TLDR
This work revisits the formulation of sparsely annotated object detection and proposes a region-based semi-supervised algorithm, that automatically identifies regions containing unlabeled foreground objects and achieves competitive performance on standard semi- supervised setups demonstrating the strength and broad applicability of the approach.
Training Object Detectors if Only Large Objects are Labeled
TLDR
This paper proposes PCIS, a novel combination of P seudo-labels, output C onsistency across scales, and an anchor scale-dependent I gnore S trategy which outperforms existing pseudo-label generation methods as well as an oracle which ensures that anchors overlapping missing annotations are ignored during training.
Semi-Supervised Object Detection With Sparsely Annotated Dataset
TLDR
The proposed technique performs bidirectional single-object tracking from sparsely annotated bounding boxes as starting points in videos to obtain dense annotations and achieves runner-up performance in the Unseen section of the EPIC-KITCHENS 2020 object detection challenge under IoU > 0.5.
Large-Scale Object Detection in the Wild From Imbalanced Multi-Labels
TLDR
This work designs a concurrent softmax to handle the multi-label problems in object detection and proposes a soft-sampling methods with hybrid training scheduler to deal with the label imbalance.
Solving Missing-Annotation Object Detection with Background Recalibration Loss
TLDR
A superior solution called Background Recalibration Loss (BRL) that can automatically re-calibrate the loss signals according to the pre-defined IoU threshold and input image and is built on the one-stage detector which is faster and lighter.
Co-mining: Self-Supervised Learning for Sparsely Annotated Object Detection
TLDR
This paper proposes a simple but effective mechanism, called Co-mining, for sparsely annotated object detection, where two branches of a Siamese network predict the pseudo-label sets for each other in order to enhance multi-view learning and better mine unlabeled instances.
SS3D: Sparsely-Supervised 3D Object Detection from Point Cloud
TLDR
The proposed SS3D framework is a general framework that can be used to learn any modern 3D object detector and outperforms the state-of-the-art weakly-supervised method by remarkable margins, highlighting its effectiveness.
A Novel Loss Calibration Strategy for Object Detection Networks Training on Sparsely Annotated Pathological Datasets
TLDR
The novel Boxes Density Energy (BDE) focuses on utilizing the densities of regression boxes to conduct loss-calibration, which is dedicated to reducing the miscalculated loss, meanwhile to penalizing mispredictions with a relatively more significant loss.
Scaling up instance annotation via label propagation
TLDR
This paper proposes a highly efficient annotation scheme for building large datasets with object segmentation masks by using hierarchical clustering on mask predictions made by a segmentation model and efficiently searches through the hierarchy of clusters and selects which clusters to annotate.
Learning an Efficient Network for Large-Scale Hierarchical Object Detection with Data Imbalance: 3rd Place Solution to Open Images Challenge 2019
TLDR
This report details the solution to the Google AI Open Images Challenge 2019 Object Detection Track, and finds that there are four typical features: large-scale, hierarchical tag system, severe annotation incompleteness and data imbalance.
...
...

References

SHOWING 1-10 OF 29 REFERENCES
Soft Sampling for Robust Object Detection
TLDR
The robustness of object detection under the presence of missing annotations is studied, and it is observed that after dropping 30% of the annotations, the performance of CNN-based object detectors like Faster-RCNN only drops by 5% on the PASCAL VOC dataset.
Training Region-Based Object Detectors with Online Hard Example Mining
TLDR
OHEM is a simple and intuitive algorithm that eliminates several heuristics and hyperparameters in common use that leads to consistent and significant boosts in detection performance on benchmarks like PASCAL VOC 2007 and 2012.
Focal Loss for Dense Object Detection
TLDR
This paper proposes to address the extreme foreground-background class imbalance encountered during training of dense detectors by reshaping the standard cross entropy loss such that it down-weights the loss assigned to well-classified examples, and develops a novel Focal Loss, which focuses training on a sparse set of hard examples and prevents the vast number of easy negatives from overwhelming the detector during training.
Solution for Large-Scale Hierarchical Object Detection Datasets with Incomplete Annotation and Data Imbalance
TLDR
This report demonstrates the solution for the Open Images 2018 Challenge using the powerful SENet154 armed with feature pyramid module and deformable ROIalign as the backbone, and finds that the best single model could achieve a mAP of 56.9%.
You Only Look Once: Unified, Real-Time Object Detection
TLDR
Compared to state-of-the-art detection systems, YOLO makes more localization errors but is less likely to predict false positives on background, and outperforms other detection methods, including DPM and R-CNN, when generalizing from natural images to other domains like artwork.
Weakly- and Semi-Supervised Object Detection with Expectation-Maximization Algorithm
TLDR
This paper develops an Expectation-Maximization (EM) based object detection method using deep convolutional neural networks (CNNs) that can almost match the performace of the fully supervised Fast RCNN.
Microsoft COCO: Common Objects in Context
We present a new dataset with the goal of advancing the state-of-the-art in object recognition by placing the question of object recognition in the context of the broader question of scene
Co-localization in Real-World Images
TLDR
An extensive evaluation of the method compared to previous state-of-the-art approaches on the challenging PASCAL VOC 2007 and Object Discovery datasets and a large-scale study of co-localization on ImageNet, involving ground-truth annotations for 3, 624 classes and approximately 1 million images.
Transfer Learning by Ranking for Weakly Supervised Object Annotation
TLDR
A novel transfer learning based on learning to rank is formulated, which effectively transfers a model for automatic annotation of object location from an auxiliary dataset to a target dataset with completely unrelated object categories.
Object Detection Meets Knowledge Graphs
TLDR
A novel framework of knowledge-aware object detection is proposed, which enables the integration of external knowledge such as knowledge graphs into any object detection algorithm, which improves object detection through a re-optimization process to achieve better consistency with background knowledge.
...
...