A Simple Semi-Supervised Learning Framework for Object Detection (2020-05-10T00:00:00.000000Z)

TL;DR

STAC is proposed, a simple yet effective SSL framework for visual object detection along with a data augmentation strategy that deploys highly confident pseudo labels of localized objects from an unlabeled image and updates the model by enforcing consistency via strong augmentations.

Abstract

Semi-supervised learning (SSL) has promising potential for improving the predictive performance of machine learning models using unlabeled data. There has been remarkable progress, but the scope of demonstration in SSL has been limited to image classification tasks. In this paper, we propose STAC, a simple yet effective SSL framework for visual object detection along with a data augmentation strategy. STAC deploys highly confident pseudo labels of localized objects from an unlabeled image and updates the model by enforcing consistency via strong augmentations. We propose new experimental protocols to evaluate performance of semi-supervised object detection using MS-COCO and demonstrate the efficacy of STAC on both MS-COCO and VOC07. On VOC07, STAC improves the AP$^{0.5}$ from 76.30 to 79.08; on MS-COCO, STAC demonstrates 2x higher data efficiency by achieving 24.38 mAP using only 5% labeled data than supervised baseline that marks 23.86% using 10% labeled data. The code is available at \url{this https URL}.

Authors

Kihyuk Sohn

8 papers

Tomas Pfister

7 papers

Chen-Yu Lee

3 papers

TL;DR

Abstract

Authors

References72 items

ReMixMatch: Semi-Supervised Learning with Distribution Matching and Augmentation Anchoring

FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence

Proposal Learning for Semi-Supervised Object Detection

SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization

AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty

EfficientDet: Scalable and Efficient Object Detection

Self-Training With Noisy Student Improves ImageNet Classification

Randaugment: Practical automated data augmentation with a reduced search space

RandAugment: Practical data augmentation with no separate search

Pseudo-Labeling and Confirmation Bias in Deep Semi-Supervised Learning

Learning Data Augmentation Strategies for Object Detection

AutoAugment: Learning Augmentation Strategies From Data

Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules

CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features

S4L: Self-Supervised Semi-Supervised Learning

MixMatch: A Holistic Approach to Semi-Supervised Learning

Unsupervised Data Augmentation for Consistency Training

Automatic Adaptation of Object Detectors to New Domains Using Self-Training

DeceptionNet: Network-Driven Domain Randomization

Bag of Freebies for Training Object Detection Neural Networks

CornerNet: Detecting Objects as Paired Keypoints

Towards Human-Machine Cooperation: Self-Supervised Sample Mining for Object Detection

Data Distillation: Towards Omni-Supervised Learning

Cascade R-CNN: Delving Into High Quality Object Detection

CyCADA: Cycle-Consistent Adversarial Domain Adaptation

mixup: Beyond Empirical Risk Minimization

Random Erasing Data Augmentation

Improved Regularization of Convolutional Neural Networks with Cutout

Focal Loss for Dense Object Detection

Cut, Paste and Learn: Surprisingly Easy Synthesis for Instance Detection

Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning

Mask R-CNN

Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results

YOLO9000: Better, Faster, Stronger

Feature Pyramid Networks for Object Detection

Temporal Ensembling for Semi-Supervised Learning

Regularization With Stochastic Transformations and Perturbations for Deep Semi-Supervised Learning

Mutual exclusivity loss for semi-supervised deep learning

R-FCN: Object Detection via Region-based Fully Convolutional Networks

Training Region-Based Object Detectors with Online Hard Example Mining

Deep Networks with Stochastic Depth

Enhanced semi-supervised learning for multimodal emotion recognition

Deep Residual Learning for Image Recognition

SSD: Single Shot MultiBox Detector

Semi-supervised Learning with Ladder Networks

You Only Look Once: Unified, Real-Time Object Detection

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Watch and learn: Semi-supervised learning of object detectors from videos

Fast R-CNN

Learning with Pseudo-Ensembles

Microsoft COCO: Common Objects in Context

Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation

ImageNet classification with deep convolutional neural networks

SUN database: Large-scale scene recognition from abbey to zoo

ImageNet: A large-scale hierarchical image database

Effective Self-Training for Parsing

Semi-Supervised Self-Training of Object Detection Models

Semi-supervised Learning by Entropy Minimization

Best practices for convolutional neural networks applied to visual document analysis

Unsupervised Word Sense Disambiguation Rivaling Supervised Methods

Probability of error of some adaptive pattern-recognition machines

Consistency-based Semi-supervised Learning for Object detection

Realistic Evaluation of Semi-Supervised Learning Algorithms

Unpaired image-to-image translation using cycle-consistent adversarial networks

Tensorpack

Large scale semi-supervised object detection using visual and semantic knowledge transfer

Dropout: a simple way to prevent neural networks from overfitting

Pseudo-Label : The Simple and Efficient Semi-Supervised Learning Method for Deep Neural Networks

Iterative Reclassification Procedure for Constructing An Asymptotically Optimal Rule of Allocation in Discriminant-Analysis

STAC is simple and introduces only two new hyperparameters: the conﬁ-dence threshold τ and the unsupervised loss weight λ u , which do not require an extensive additional eﬀort for tuning

We propose new experimental protocols for SSL object detection using MS-COCO and demonstrate the eﬃcacy of STAC on MS-COCO and PASCAL VOC in Faster RCNN framework

International Journal of Computer Vision manuscript No. (will be inserted by the editor) The PASCAL Visual Object Classes (VOC) Challenge

Field of Study

Journal Information

Name

Volume