Pattern Recognition and Computer Vision : 6th Chinese Conference, PRCV 2023, Xiamen, China, October 13–15, 2023, Proceedings, Part XII / edited by Qingshan Liu, Hanzi Wang, Zhanyu Ma, Weishi Zheng, Hongbin Zha, Xilin Chen, Liang Wang, Rongrong Ji.

Format:

Book

Author/Creator:

Liu, Qingshan, -1952, author.

Series:

Lecture Notes in Computer Science, 1611-3349 ; 14436

Language:

English

Subjects (All):

Image processing--Digital techniques.

Image processing.

Computer vision.

Artificial intelligence.

Application software.

Computer networks.

Computer systems.

Machine learning.

Computer Imaging, Vision, Pattern Recognition and Graphics.

Artificial Intelligence.

Computer and Information Systems Applications.

Computer Communication Networks.

Computer System Implementation.

Machine Learning.

Local Subjects:

Computer Imaging, Vision, Pattern Recognition and Graphics.

Artificial Intelligence.

Computer and Information Systems Applications.

Computer Communication Networks.

Computer System Implementation.

Machine Learning.

Physical Description:

1 online resource (XIV, 523 p. 203 illus., 194 illus. in color.)

Edition:

1st ed. 2024.

Place of Publication:

Singapore : Springer Nature Singapore : Imprint: Springer, 2024.

Summary:

The 13-volume set LNCS 14425-14437 constitutes the refereed proceedings of the 6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023, held in Xiamen, China, during October 13–15, 2023. The 532 full papers presented in these volumes were selected from 1420 submissions. The papers have been organized in the following topical sections: Action Recognition, Multi-Modal Information Processing, 3D Vision and Reconstruction, Character Recognition, Fundamental Theory of Computer Vision, Machine Learning, Vision Problems in Robotics, Autonomous Driving, Pattern Classification and Cluster Analysis, Performance Evaluation and Benchmarks, Remote Sensing Image Interpretation, Biometric Recognition, Face Recognition and Pose Recognition, Structural Pattern Recognition, Computational Photography, Sensing and Display Technology, Video Analysis and Understanding, Vision Applications and Systems, Document Analysis and Recognition, Feature Extraction and Feature Selection, Multimedia Analysis and Reasoning, Optimization and Learning methods, Neural Network and Deep Learning, Low-Level Vision and Image Processing, Object Detection, Tracking and Identification, Medical Image Processing and Analysis. .

Contents:

Intro

Preface

Organization

Contents - Part XII

Object Detection, Tracking and Identification

OKGR: Occluded Keypoint Generation and Refinement for 3D Object Detection

1 Introduction

2 Related Works

2.1 LiDAR-Based 3D Object Detection

2.2 Object Shape Completion

3 Methodology

3.1 Overview

3.2 Occluded Keypoint Generation

3.3 Occluded Keypoint Refinement

3.4 Loss Function

4 Experiments

4.1 Datasets and Evaluation Metrics

4.2 Implementation Details

4.3 Evaluation on KITTI Dataset

4.4 Evaluation on Waymo Open Dataset

4.5 Model Efficiency

4.6 Ablation Studies

5 Conclusion

References

Camouflaged Object Segmentation Based on Fractional Edge Perception

2 Related Work

3 Interactive Task Learning Network

3.1 Integral and Fractional Edge

3.2 Camouflaged Edge Detection Module

4 Performance Evaluation

4.1 Datasets and Experiment Settings

4.2 Quantitative Evaluation

4.3 Qualitative Evaluation

4.4 Generalization of Edge Detection

DecTrans: Person Re-identification with Multifaceted Part Features via Decomposed Transformer

3.1 Vision Transformer as Feature Extractor

3.2 Token Decomposition (TD) Layer

3.3 Data Augmentation for TD Layer

3.4 Training and Inference

4.3 Comparisons to State-of-the-arts

4.4 Ablation Study

AHT: A Novel Aggregation Hyper-transformer for Few-Shot Object Detection

2.1 Object Detection

2.2 Hypernetworks

3 Method

3.1 Preliminaries

3.2 Overview

3.3 Dynamic Aggregation Module

3.4 Conditional Adaptation Hypernetworks.

3.5 The Classification-Regression Detection Head

4.1 Experimental Setting

4.2 Comparison Results

4.3 Ablation Study

4.4 Visualization of Our Module

Feature Refinement from Multiple Perspectives for High Performance Salient Object Detection

2 Proposed Method

2.1 Overall Architecture

2.2 Attention-Guided Bi-directional Feature Refinement Module

2.3 Serial Atrous Fusion Module

2.4 Upsampling Feature Refinement Module

2.5 Objective Function

3 Experiments

3.1 Experimental Setup

3.2 Comparison with State-of-the-Art Methods

3.3 Ablation Study

4 Conclusion

Feature Disentanglement and Adaptive Fusion for Improving Multi-modal Tracking

2.1 Multi-modal Tracking

2.2 Transformers Tracking

3.1 Preliminary

3.2 Our Approach

3.3 Training and Inference

4.1 Implementation Details

4.2 Comparison with State-of-the-Arts Multi-modal Trackers

Modality Balancing Mechanism for RGB-Infrared Object Detection in Aerial Image

2.1 Object Detection in Aerial Images

2.2 RGB-Infrared Object Detection

3.2 Modality Balancing Mechanism

3.3 Multimodal Feature Hybrid Sampling Module

4 Experiment

4.1 Settings

4.2 Comparison with State-of-the-Art Methods

Pacific Oyster Gonad Identification and Grayscale Calculation Based on Unapparent Object Detection

2 Method

2.1 Compact Pyramid Refinement Module (CPRM)

2.2 Switchable Excitation Model (SEM)

3 Experiments and Analysis of Results

3.1 Establishment of the Datasets.

3.2 Experimental Environment and Evaluation Index

3.3 Ablation Experiments

3.4 Comparative Experiments and Analysis of Results

3.5 Visualization Results

3.6 Gray Value Calculation

Multi-task Self-supervised Few-Shot Detection

2.1 Self-supervised Learning

2.2 Few-Shot Object Detection

3.1 Problem Setting

3.2 Self-supervised Auxiliary Branch

3.3 Multi-Task Learning

4.2 Few-Shot Object Detection Benchmarks

4.3 Ablation Analysis

4.4 Visualization

CSTrack: A Comprehensive and Concise Vision Transformer Tracker

3.2 CSBlock

3.3 Prediction Head and Loss

4.2 Comparisons with the State-of-the-Art Trackers

4.4 Visualization of Attention Maps

4.5 Visualization of Tracking Performance

Feature Implicit Enhancement via Super-Resolution for Small Object Detection

2.1 General Object Detection

2.2 Small Object Detection Based on Super-Resolution

3 Methods

3.1 Overall Architecture

3.2 Training

4 Experiments and Details

4.1 Dataset and Details

4.2 Ablation Study

4.3 Main Results

Improved Detection Method for SODL-YOLOv7 Intensive Juvenile Abalone

2 Methods

2.1 SODL Small Target Detection Network

2.2 ACBAM Attention Module

3 Experimental Results and Analysis

3.1 Experimental Data Preprocessing

3.2 Experimental Environment and Evaluation Index

3.3 Experimental Results and Analysis

References.

MVP-SEG: Multi-view Prompt Learning for Open-Vocabulary Semantic Segmentation

2.1 Vision-Language Models

2.2 Zero-Shot Segmentation

2.3 Prompt Learning

3.1 Problem Definition

3.2 MVP-SEG

3.3 MVP-SEG+

4.1 Datasets

4.2 Evaluation Metrics

4.3 Implementation Details

4.4 Ablation Studies on MVP-SEG

4.5 Comparison with State-of-the-Art

Context-FPN and Memory Contrastive Learning for Partially Supervised Instance Segmentation

3 CCMask

3.2 Context-FPN

3.3 Memory Contrastive Learning Head

4.1 Experimental Setup

4.2 Experimental Results

A Dynamic Tracking Framework Based on Scene Perception

3.1 Easy-Hard Dual-Branch Network

3.2 Scene Router

4.2 Comparison with State-of-the-arts

4.3 Ablation Study and Analysis

HPAN: A Hybrid Pose Attention Network for Person Re-Identification

2 The Proposed Method

2.1 Local Key Point Features

2.2 Self-Attention

2.3 Hybrid Pose and Global Feature Fusion (HPGFF)

2.4 Loss Function

2.5 Training Strategy

3.1 Datasets and Evaluation Metrics

3.2 Comparison with SOTA Methods

3.3 Ablation Studies

3.4 Visualization of Attention Maps

SpectralTracker: Jointly High and Low-Frequency Modeling for Tracking

2.1 Visual Tracking

2.2 Frequency Modeling in Visual Transformer

3.1 Dual-Spectral Module

3.2 Dual-Spectral for Tracking

3.3 Prediction Head and Total Loss.

4 Experiments

4.2 State-of-the-Art Comparison

4.3 Ablation Studies

DiffusionTracker: Targets Denoising Based on Diffusion Model for Visual Tracking

2.1 Visual Tracking Based on Siamese Network

2.2 Diffusion Model

3.1 Architecture

3.2 Training Process

3.3 Inference Process

4.3 General Datasets Evaluation

4.4 Attributes Evaluation

4.5 Compatibility Experiment

Instance-Proxy Loss for Semi-supervised Learning with Coarse Labels

3.1 Instance-Level Loss

3.2 Proxy-Level Loss

3.3 Instance-Proxy Loss

4.1 Comparison to SOTA Methods

FAFVTC: A Real-Time Network for Vehicle Tracking and Counting

3.1 Backbone Network

3.2 Multi-spectral Channel and Spatial Attention (MCSA)

3.3 Data Association

3.4 Vehicle Counting

4.1 Datasets and Metrics

4.3 Comparison Experiments

Ped-Mix: Mix Pedestrians for Occluded Person Re-identification

2.1 Occluded Person Re-identification

2.2 Data Augmentation and Training Loss

3 Proposed Method

3.1 Ped-Mix

3.2 Non-target Suppression Loss

3.3 Training Procedure

4.1 Datasets and Evaluation Measures

4.4 Comparison with State-of-the-Art Methods

4.5 Visualization

4.6 Why Random Masking

4.7 Results on Holistic Datasets

Object-Aware Transfer-Based Black-Box Adversarial Attack on Object Detector.

ISBN:

981-9985-55-2