Path Aggregation Network for Instance Segmentation

Abstract

The way that information propagates in neural networks is of great importance. In this paper, we propose Path Aggregation Network (PANet) aiming at boosting information flow in proposal-based instance segmentation framework. Specifically, we enhance the entire feature hierarchy with accurate localization signals in lower layers by bottom-up path augmentation, which shortens the information path between lower layers and topmost feature. We present adaptive feature pooling, which links feature grid and all feature levels to make useful information in each feature level propagate directly to following proposal subnetworks. A complementary branch capturing different views for each proposal is created to further improve mask prediction. These improvements are simple to implement, with subtle extra computational overhead. Our PANet reaches the 1st place in the COCO 2017 Challenge Instance Segmentation task and the 2nd place in Object Detection task without large-batch training. It is also state-of-the-art on MVD and Cityscapes. Code is available at https://github.com/ShuLiu1993/PANet

arXiv commentAccepted to CVPR 2018

References in corpus (14)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Caffe: Convolutional Architecture for Fast Feature Embedding
• Fully Convolutional Networks for Semantic Segmentation
• Squeeze-and-Excitation Networks
• DSSD : Deconvolutional Single Shot Detector
• ParseNet: Looking Wider to See Better
• Learning to Segment Object Candidates
• Convolutional Feature Masking for Joint Object and Stuff Segmentation
• Beyond Skip Connections: Top-Down Modulation for Object Detection
• Learning to Refine Object Segments
• Deep Watershed Transform for Instance Segmentation
• Dual Path Networks
• Crafting GBD-Net for Object Detection

Cited by in corpus (52)

• Learning Spatial Fusion for Single-Shot Object Detection
• Evolution of Image Segmentation using Deep Convolutional Neural Network: A Survey
• Pyramid Attention Network for Semantic Segmentation
• Context Encoding for Semantic Segmentation
• Deblending and Classifying Astronomical Sources with Mask R-CNN Deep Learning
• Rethinking ImageNet Pre-training
• Retina U-Net: Embarrassingly Simple Exploitation of Segmentation Supervision for Medical Object Detection
• 3D RoI-aware U-Net for Accurate and Efficient Colorectal Tumor Segmentation
• Panoptic Segmentation with a Joint Semantic and Instance Segmentation Network
• One-Shot Instance Segmentation
• Prime Sample Attention in Object Detection
• AugFPN: Improving Multi-scale Feature Learning for Object Detection
• Learning Gaussian Instance Segmentation in Point Clouds
• PointGroup: Dual-Set Point Grouping for 3D Instance Segmentation
• Instance Segmentation by Deep Coloring
• CentripetalNet: Pursuing High-quality Keypoint Pairs for Object Detection
• FoodLogoDet-1500: A Dataset for Large-Scale Food Logo Detection via Multi-Scale Feature Decoupling Network
• Squeeze-and-Attention Networks for Semantic Segmentation
• Side-Aware Boundary Localization for More Precise Object Detection
• Solution for Large-Scale Hierarchical Object Detection Datasets with Incomplete Annotation and Data Imbalance
• Hit-Detector: Hierarchical Trinity Architecture Search for Object Detection
• PolyTransform: Deep Polygon Transformer for Instance Segmentation
• Fully Convolutional Networks for Panoptic Segmentation
• BoundarySqueeze: Image Segmentation as Boundary Squeezing
• Spatial Feature Calibration and Temporal Fusion for Effective One-stage Video Instance Segmentation
• IPG-Net: Image Pyramid Guidance Network for Small Object Detection
• 1st Place Solutions of Waymo Open Dataset Challenge 2020 -- 2D Object Detection Track
• Instance Shadow Detection
• Rethinking Training from Scratch for Object Detection
• Topological Map Extraction from Overhead Images
• IDD: A Dataset for Exploring Problems of Autonomous Navigation in Unconstrained Environments
• Instance and Panoptic Segmentation Using Conditional Convolutions
• Outline Objects using Deep Reinforcement Learning
• Deep Feature Pyramid Reconfiguration for Object Detection
• Reducing Label Noise in Anchor-Free Object Detection
• Effective Fusion Factor in FPN for Tiny Object Detection
• Bipartite Conditional Random Fields for Panoptic Segmentation
• Pedestrian Detection with Autoregressive Network Phases
• Parsing R-CNN for Instance-Level Human Analysis
• HR-RCNN: Hierarchical Relational Reasoning for Object Detection
• Weakly- and Semi-Supervised Panoptic Segmentation
• CASNet: Common Attribute Support Network for image instance and panoptic segmentation
• OVSNet : Towards One-Pass Real-Time Video Object Segmentation
• Histogram of Cell Types: Deep Learning for Automated Bone Marrow Cytology
• Efficient Human Pose Estimation by Learning Deeply Aggregated Representations
• Universal-RCNN: Universal Object Detector via Transferable Graph R-CNN
• Efficient Video Understanding via Layered Multi Frame-Rate Analysis
• SparseMask: Differentiable Connectivity Learning for Dense Image Prediction
• Implicit Label Augmentation on Partially Annotated Clips via Temporally-Adaptive Features Learning
• ContourRender: Detecting Arbitrary Contour Shape For Instance Segmentation In One Pass
• 3rd Place Scheme on Instance Segmentation Track of ICCV 2021 VIPriors Challenges
• Accurate Nuclear Segmentation with Center Vector Encoding