MMDetection: Open MMLab Detection Toolbox and Benchmark
arXiv:1906.07155
Abstract
We present MMDetection, an object detection toolbox that contains a rich set of object detection and instance segmentation methods as well as related components and modules. The toolbox started from a codebase of MMDet team who won the detection track of COCO Challenge 2018. It gradually evolves into a unified platform that covers many popular detection methods and contemporary modules. It not only includes training and inference codes, but also provides weights for more than 200 network models. We believe this toolbox is by far the most complete detection toolbox. In this paper, we introduce the various features of this toolbox. In addition, we also conduct a benchmarking study on different methods, components, and their hyper-parameters. We wish that the toolbox and benchmark could serve the growing research community by providing a flexible toolkit to reimplement existing methods and develop their own new detectors. Code and models are available at https://github.com/open-mmlab/mmdetection. The project is under active development and we will keep this document updated.
Technical report of MMDetection. 11 pages
References in corpus (13)
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- R-FCN: Object Detection via Region-based Fully Convolutional Networks
- Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
- MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems
- FCOS: Fully Convolutional One-Stage Object Detection
- High-Resolution Representations for Labeling Pixels and Regions
- GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond
- Micro-Batch Training with Batch-Channel Normalization and Weight Standardization
- An Empirical Study of Spatial Attention Mechanisms in Deep Networks
- Rethinking ImageNet Pre-training
- Scale-Aware Trident Networks for Object Detection
- Rethinking Classification and Localization for Object Detection
- SimpleDet: A Simple and Versatile Distributed Framework for Object Detection and Instance Recognition
Cited by in corpus (279)
- PVT v2: Improved Baselines with Pyramid Vision Transformer
- Transformer in Transformer
- Object Detection in Aerial Images: A Large-Scale Benchmark and Challenges
- SOLOv2: Dynamic and Fast Instance Segmentation
- Prototypical Contrastive Learning of Unsupervised Representations
- Anchor-free Oriented Proposal Generator for Object Detection
- ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks
- Deep High-Resolution Representation Learning for Visual Recognition
- HIT-UAV: A high-altitude infrared thermal dataset for Unmanned Aerial Vehicle-based object detection
- Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection
- A Normalized Gaussian Wasserstein Distance for Tiny Object Detection
- A Dataset And Benchmark Of Underwater Object Detection For Robot Picking
- XCiT: Cross-Covariance Image Transformers
- K-Net: Towards Unified Image Segmentation
- GhostNets on Heterogeneous Devices via Cheap Operations
- Probabilistic two-stage detection
- DetectoRS: Detecting Objects with Recursive Feature Pyramid and Switchable Atrous Convolution
- Alpha-IoU: A Family of Power Intersection over Union Losses for Bounding Box Regression
- Benchmarking Robustness in Object Detection: Autonomous Driving when Winter is Coming
- SSPNet: Scale Selection Pyramid Network for Tiny Person Detection from UAV Images
- Fewer is More: Efficient Object Detection in Large Aerial Images
- Reinventing 2D Convolutions for 3D Images
- CSWin Transformer: A General Vision Transformer Backbone with Cross-Shaped Windows
- CycleMLP: A MLP-like Architecture for Dense Prediction
- Torchreid: A Library for Deep Learning Person Re-Identification in Pytorch
- Attention-Based Transformers for Instance Segmentation of Cells in Microstructures
- A novel Region of Interest Extraction Layer for Instance Segmentation
- Beyond Self-attention: External Attention using Two Linear Layers for Visual Tasks
- InsPLAD: A Dataset and Benchmark for Power Line Asset Inspection in UAV Images
- Learning N:M Fine-grained Structured Sparse Neural Networks From Scratch
- Sample and Computation Redistribution for Efficient Face Detection
- RepPoints V2: Verification Meets Regression for Object Detection
- Optimizing Rank-based Metrics with Blackbox Differentiation
- Asymmetric Loss For Multi-Label Classification
- PolarMask: Single Shot Instance Segmentation with Polar Representation
- Aligning Pretraining for Detection via Object-Level Contrastive Learning
- Rotate to Attend: Convolutional Triplet Attention Module
- Uncertainty for Identifying Open-Set Errors in Visual Object Detection
- Long-Tailed Classification by Keeping the Good and Removing the Bad Momentum Causal Effect
- Joint Monocular 3D Vehicle Detection and Tracking
- Reference-based Defect Detection Network
- CARAFE: Content-Aware ReAssembly of FEatures
- Prime Sample Attention in Object Detection
- AtomNAS: Fine-Grained End-to-End Neural Architecture Search
- ReDet: A Rotation-equivariant Detector for Aerial Object Detection
- DADA: Differentiable Automatic Data Augmentation
- Object Relational Graph with Teacher-Recommended Learning for Video Captioning
- ResizeMix: Mixing Data with Preserved Object Information and True Labels
- FcaNet: Frequency Channel Attention Networks
- YOLO5Face: Why Reinventing a Face Detector
- DenseCLIP: Language-Guided Dense Prediction with Context-Aware Prompting
- Noisy Differentiable Architecture Search
- RelationNet++: Bridging Visual Representations for Object Detection via Transformer Decoder
- Distilling Object Detectors with Feature Richness
- Glance-and-Gaze Vision Transformer
- VarifocalNet: An IoU-aware Dense Object Detector
- Gradient Centralization: A New Optimization Technique for Deep Neural Networks
- PeCo: Perceptual Codebook for BERT Pre-training of Vision Transformers
- Generation of microbial colonies dataset with deep learning style transfer
- MetaFormer Is Actually What You Need for Vision
- Fair DARTS: Eliminating Unfair Advantages in Differentiable Architecture Search
- DARTS-: Robustly Stepping out of Performance Collapse Without Indicators
- FoodLogoDet-1500: A Dataset for Large-Scale Food Logo Detection via Multi-Scale Feature Decoupling Network
- On Model Calibration for Long-Tailed Object Detection and Instance Segmentation
- A Ranking-based, Balanced Loss Function Unifying Classification and Localisation in Object Detection
- Instant-Teaching: An End-to-End Semi-Supervised Object Detection Framework
- Distilling Object Detectors via Decoupled Features
- Exploring Data Augmentation for Multi-Modality 3D Object Detection
- ConvMLP: Hierarchical Convolutional MLPs for Vision
- ReMOTS: Self-Supervised Refining Multi-Object Tracking and Segmentation
- Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection
- Corner Proposal Network for Anchor-free, Two-stage Object Detection
- GradAug: A New Regularization Method for Deep Neural Networks
- Revisiting Locally Supervised Learning: an Alternative to End-to-end Training
- Propagate Yourself: Exploring Pixel-Level Consistency for Unsupervised Visual Representation Learning
- Regularizing Deep Networks with Semantic Data Augmentation
- Semantic Relation Reasoning for Shot-Stable Few-Shot Object Detection
- Focal and Global Knowledge Distillation for Detectors
- See Eye to Eye: A Lidar-Agnostic 3D Detection Framework for Unsupervised Multi-Target Domain Adaptation
- Side-Aware Boundary Localization for More Precise Object Detection
- Empirical Upper Bound in Object Detection and More
- LoCo: Local Contrastive Representation Learning
- Anchor Pruning for Object Detection
- Visual Parser: Representing Part-whole Hierarchies with Transformers
- Adaptable Deformable Convolutions for Semantic Segmentation of Fisheye Images in Autonomous Driving Systems
- TOOD: Task-aligned One-stage Object Detection
- Loss Function Discovery for Object Detection via Convergence-Simulation Driven Search
- HRDNet: High-resolution Detection Network for Small Objects
- RoIMix: Proposal-Fusion among Multiple Images for Underwater Object Detection
- RDSNet: A New Deep Architecture for Reciprocal Object Detection and Instance Segmentation
- FCOSR: A Simple Anchor-free Rotated Detector for Aerial Object Detection
- Analyzing Worldwide Social Distancing through Large-Scale Computer Vision
- CompFeat: Comprehensive Feature Aggregation for Video Instance Segmentation
- Towards Noise-resistant Object Detection with Noisy Annotations
- Align Deep Features for Oriented Object Detection
- MixPath: A Unified Approach for One-shot Neural Architecture Search
- IoU-aware Single-stage Object Detector for Accurate Localization
- Densely Connected Search Space for More Flexible Neural Architecture Search
- MovieNet: A Holistic Dataset for Movie Understanding
- A Competitive Method to VIPriors Object Detection Challenge
- 1st Place Solution to ECCV-TAO-2020: Detect and Represent Any Object for Tracking
- LF-YOLO: A Lighter and Faster YOLO for Weld Defect Detection of X-ray Image
- BANet: Bidirectional Aggregation Network with Occlusion Handling for Panoptic Segmentation
- Look-into-Object: Self-supervised Structure Modeling for Object Recognition
- 1st Place Solutions for OpenImage2019 -- Object Detection and Instance Segmentation
- Recurrence along Depth: Deep Convolutional Neural Networks with Recurrent Layer Aggregation
- Soft Anchor-Point Object Detection
- UFO-ViT: High Performance Linear Vision Transformer without Softmax
- MMFashion: An Open-Source Toolbox for Visual Fashion Analysis
- Proposal Learning for Semi-Supervised Object Detection
- MSG-Transformer: Exchanging Local Spatial Information by Manipulating Messenger Tokens
- SpatialFlow: Bridging All Tasks for Panoptic Segmentation
- Disentangled Non-Local Neural Networks
- Deep neural networks approach to microbial colony detection -- a comparative analysis
- FAIR1M: A Benchmark Dataset for Fine-grained Object Recognition in High-Resolution Remote Sensing Imagery
- ActNN: Reducing Training Memory Footprint via 2-Bit Activation Compressed Training
- DCT-Mask: Discrete Cosine Transform Mask Representation for Instance Segmentation
- IoU-uniform R-CNN: Breaking Through the Limitations of RPN
- Equalization Loss v2: A New Gradient Balance Approach for Long-tailed Object Detection
- Combating Noise: Semi-supervised Learning by Region Uncertainty Quantification
- Wildfire Smoke Detection System: Model Architecture, Training Mechanism, and Dataset
- PolarMask++: Enhanced Polar Representation for Single-Shot Instance Segmentation and Beyond
- Classification Calibration for Long-tail Instance Segmentation
- Overcoming Classifier Imbalance for Long-tail Object Detection with Balanced Group Softmax
- CARAFE++: Unified Content-Aware ReAssembly of FEatures
- CAT: Cross-Attention Transformer for One-Shot Object Detection
- Rethinking Normalization and Elimination Singularity in Neural Networks
- Local Metrics for Multi-Object Tracking
- MultiSports: A Multi-Person Video Dataset of Spatio-Temporally Localized Sports Actions
- PingAn-VCGroup's Solution for ICDAR 2021 Competition on Scientific Table Image Recognition to Latex
- DCANet: Learning Connected Attentions for Convolutional Neural Networks
- Scale-Equalizing Pyramid Convolution for Object Detection
- 1st Place Solutions of Waymo Open Dataset Challenge 2020 -- 2D Object Detection Track
- Rethinking Natural Adversarial Examples for Classification Models
- Unmanned Aerial Vehicle Visual Detection and Tracking using Deep Neural Networks: A Performance Benchmark
- hSDB-instrument: Instrument Localization Database for Laparoscopic and Robotic Surgeries
- FASA: Feature Augmentation and Sampling Adaptation for Long-Tailed Instance Segmentation
- Perception for Autonomous Systems (PAZ)
- Localization Distillation for Dense Object Detection
- Hire-MLP: Vision MLP via Hierarchical Rearrangement
- Rethinking Training from Scratch for Object Detection
- SWA Object Detection
- 2nd Place Solution for Waymo Open Dataset Challenge -- 2D Object Detection
- FC2RN: A Fully Convolutional Corner Refinement Network for Accurate Multi-Oriented Scene Text Detection
- Towards large-scale, automated, accurate detection of CCTV camera objects using computer vision. Applications and implications for privacy, safety, and cybersecurity. (Preprint)
- 1st Place Solution for ICDAR 2021 Competition on Mathematical Formula Detection
- Learning to Exploit Multiple Vision Modalities by Using Grafted Networks
- Rank & Sort Loss for Object Detection and Instance Segmentation
- Methodology for generating synthetic labeled datasets for visual container inspection
- FIBA: Frequency-Injection based Backdoor Attack in Medical Image Analysis
- 1st Place Solution for the UVO Challenge on Image-based Open-World Segmentation 2021
- Decoupled and Memory-Reinforced Networks: Towards Effective Feature Learning for One-Step Person Search
- Pseudo-IoU: Improving Label Assignment in Anchor-Free Object Detection
- Channel Equilibrium Networks for Learning Deep Representation
- DeepACC:Automate Chromosome Classification based on Metaphase Images using Deep Learning Framework Fused with Prior Knowledge
- RefineMask: Towards High-Quality Instance Segmentation with Fine-Grained Features
- Seesaw Loss for Long-Tailed Instance Segmentation
- Data-Uncertainty Guided Multi-Phase Learning for Semi-Supervised Object Detection
- DeU-Net: Deformable U-Net for 3D Cardiac MRI Video Segmentation
- AlphaRotate: A Rotation Detection Benchmark using TensorFlow
- Towards Resolving the Challenge of Long-tail Distribution in UAV Images for Object Detection
- A Solution to Product detection in Densely Packed Scenes
- CDeC-Net: Composite Deformable Cascade Network for Table Detection in Document Images
- Locally Enhanced Self-Attention: Combining Self-Attention and Convolution as Local and Context Terms
- Oriented R-CNN for Object Detection
- Improvement of Classification in One-Stage Detector
- Aggregation and Finetuning for Clothes Landmark Detection
- Density Map Guided Object Detection in Aerial Images
- Learning a Unified Sample Weighting Network for Object Detection
- FNA++: Fast Network Adaptation via Parameter Remapping and Architecture Search
- MutualNet: Adaptive ConvNet via Mutual Learning from Network Width and Resolution
- Deep Variational Instance Segmentation
- The Devil is in Classification: A Simple Framework for Long-tail Object Detection and Instance Segmentation
- Mask Encoding for Single Shot Instance Segmentation
- GLSD: The Global Large-Scale Ship Database and Baseline Evaluations
- Adaptive Class Suppression Loss for Long-Tail Object Detection
- Foreground-Background Imbalance Problem in Deep Object Detectors: A Review
- CPM R-CNN: Calibrating Point-guided Misalignment in Object Detection
- Split to Be Slim: An Overlooked Redundancy in Vanilla Convolution
- High-Performance Long-Term Tracking with Meta-Updater
- Multiple Anchor Learning for Visual Object Detection
- A Graph-Based Framework to Bridge Movies and Synopses
- Sampling Equivariant Self-attention Networks for Object Detection in Aerial Images
- Designovel's system description for Fashion-IQ challenge 2019
- MSN: Efficient Online Mask Selection Network for Video Instance Segmentation
- LIP: Local Importance-based Pooling
- ASAP-NMS: Accelerating Non-Maximum Suppression Using Spatially Aware Priors
- Noisy Annotation Refinement for Object Detection
- Social Adaptive Module for Weakly-supervised Group Activity Recognition
- A^2-FPN: Attention Aggregation based Feature Pyramid Network for Instance Segmentation
- AutoLoss-Zero: Searching Loss Functions from Scratch for Generic Tasks
- Learning to Track Instances without Video Annotations
- EagerMOT: 3D Multi-Object Tracking via Sensor Fusion
- Anchor-Free Person Search
- Understanding the Effects of Pre-Training for Object Detectors via Eigenspectrum
- Estimates of maize plant density from UAV RGB images using Faster-RCNN detection model: impact of the spatial resolution
- Augmenting Proposals by the Detector Itself
- Disentangle Your Dense Object Detector
- Towards Streaming Perception
- Direct Differentiable Augmentation Search
- 1st Place Solution for YouTubeVOS Challenge 2021:Video Instance Segmentation
- Linear Context Transform Block
- TransMix: Attend to Mix for Vision Transformers
- BA^2M: A Batch Aware Attention Module for Image Classification
- Resisting Crowd Occlusion and Hard Negatives for Pedestrian Detection in the Wild
- End-to-end Deep Learning Methods for Automated Damage Detection in Extreme Events at Various Scales
- The MIS Check-Dam Dataset for Object Detection and Instance Segmentation Tasks
- What Deep CNNs Benefit from Global Covariance Pooling: An Optimization Perspective
- MosaicOS: A Simple and Effective Use of Object-Centric Images for Long-Tailed Object Detection
- Are standard Object Segmentation models sufficient for Learning Affordance Segmentation?
- Probabilistic Ranking-Aware Ensembles for Enhanced Object Detections
- Efficient Person Search: An Anchor-Free Approach
- You Cannot Easily Catch Me: A Low-Detectable Adversarial Patch for Object Detectors
- Instance Segmentation Challenge Track Technical Report, VIPriors Workshop at ICCV 2021: Task-Specific Copy-Paste Data Augmentation Method for Instance Segmentation
- Transformer Assisted Convolutional Network for Cell Instance Segmentation
- PTQ-SL: Exploring the Sub-layerwise Post-training Quantization
- Detecting Out-of-distribution Objects Using Neuron Activation Patterns
- Batch Normalization with Enhanced Linear Transformation
- GreedyNASv2: Greedier Search with a Greedy Path Filter
- Learning Universal Shape Dictionary for Realtime Instance Segmentation
- On Cropped versus Uncropped Training Sets in Tabular Structure Detection
- Global Context Aware RCNN for Object Detection
- Discriminative Semantic Feature Pyramid Network with Guided Anchoring for Logo Detection
- RaftMLP: How Much Can Be Done Without Attention and with Less Spatial Locality?
- A Holistically-Guided Decoder for Deep Representation Learning with Applications to Semantic Segmentation and Object Detection
- Global Context Networks
- IC Networks: Remodeling the Basic Unit for Convolutional Neural Networks
- LGA-RCNN: Loss-Guided Attention for Object Detection
- Autonomous detection of molecular configurations in microscopic images based on deep convolutional neural network
- Decoupled Dynamic Filter Networks
- Learning Polar Encodings for Arbitrary-Oriented Ship Detection in SAR Images
- Variational Pedestrian Detection
- An Efficient Accelerator Design Methodology for Deformable Convolutional Networks
- Coherent Reconstruction of Multiple Humans from a Single Image
- OF-VO: Efficient Navigation among Pedestrians Using Commodity Sensors
- Feature-Driven Super-Resolution for Object Detection
- Video Panoptic Segmentation
- Neural Mesh Refiner for 6-DoF Pose Estimation
- Pseudo-Labeling for Small Lesion Detection on Diabetic Retinopathy Images
- BABO: Background Activation Black-Out for Efficient Object Detection
- PBRnet: Pyramidal Bounding Box Refinement to Improve Object Localization Accuracy
- 2nd Place Solution to ECCV 2020 VIPriors Object Detection Challenge
- 15 Keypoints Is All You Need
- ALET (Automated Labeling of Equipment and Tools): A Dataset, a Baseline and a Usecase for Tool Detection in the Wild
- Plug & Play Convolutional Regression Tracker for Video Object Detection
- IMMVP: An Efficient Daytime and Nighttime On-Road Object Detector
- Understanding the Disharmony between Weight Normalization Family and Weight Decay: shifted Regularizer
- SAIS: Single-stage Anchor-free Instance Segmentation
- Suppress-and-Refine Framework for End-to-End 3D Object Detection
- USB: Universal-Scale Object Detection Benchmark
- Modulating Localization and Classification for Harmonized Object Detection
- Recursively Refined R-CNN: Instance Segmentation with Self-RoI Rebalancing
- Searching by Generating: Flexible and Efficient One-Shot NAS with Architecture Generator
- Multi-Scale Aligned Distillation for Low-Resolution Detection
- Distilling Image Classifiers in Object Detectors
- Custom Object Detection via Multi-Camera Self-Supervised Learning
- Dynamic Normalization
- Towards Good Practices for Instance Segmentation
- Decoupled Self Attention for Accurate One Stage Object Detection
- Learning to Generate Content-Aware Dynamic Detectors
- 3rd Place Scheme on Instance Segmentation Track of ICCV 2021 VIPriors Challenges
- Accurate Fine-grained Layout Analysis for the Historical Tibetan Document Based on the Instance Segmentation
- ViP: Video Platform for PyTorch
- Semi-supervised Learning for Dense Object Detection in Retail Scenes
- DSIC: Dynamic Sample-Individualized Connector for Multi-Scale Object Detection
- UVO Challenge on Video-based Open-World Segmentation 2021: 1st Place Solution
- Poly-NL: Linear Complexity Non-local Layers with Polynomials
- Condensing Two-stage Detection with Automatic Object Key Part Discovery
- Deeply Shared Filter Bases for Parameter-Efficient Convolutional Neural Networks
- Recognition of Russian traffic signs in winter conditions. Solutions of the "Ice Vision" competition winners
- Objects as Extreme Points
- Adversarial Semantic Contour for Object Detection
- Delving into the Imbalance of Positive Proposals in Two-stage Object Detection
- CompConv: A Compact Convolution Module for Efficient Feature Learning
- Empirical Upper Bound, Error Diagnosis and Invariance Analysis of Modern Object Detectors
- ContourRender: Detecting Arbitrary Contour Shape For Instance Segmentation In One Pass
- PANDA: A Gigapixel-level Human-centric Video Dataset
- GPCA: A Probabilistic Framework for Gaussian Process Embedded Channel Attention
- Towards Fine-grained Large Object Segmentation 1st Place Solution to 3D AI Challenge 2020 -- Instance Segmentation Track