Self-supervised monocular Depth estimation with multi-scale structure similarity loss

Article Health Care Sciences & Services

Effective CBMIR System Using Hybrid Features-Based Independent Condensed Nearest Neighbor Model

Hirald Dwaraka Praveena et al.

Summary: This research studied an effective automated system for medical image retrieval by improving visible quality of input medical images and accomplishing hybrid feature extraction, achieving a retrieval accuracy of 98.88%.

JOURNAL OF HEALTHCARE ENGINEERING (2022)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

RGB-Depth Fusion GAN for Indoor Depth Completion

Haowen Wang et al.

Summary: This paper presents a novel network model for depth completion and introduces the structures and design principles of two branches. Experimental results demonstrate the effectiveness of this method in improving depth completion performance, especially in simulating realistic indoor environments.

2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022) (2022)

Add to Collection

Article Computer Science, Artificial Intelligence

Activity guided multi-scales collaboration based on scaled-CNN for saliency prediction

Deqiang Cheng et al.

Summary: This study introduces a lightweight saliency prediction model based on convolutional neural networks, utilizing multi-scale collaboration learning of global and local information, achieving competitive and consistent results on challenging benchmark datasets with better prediction performance, fewer parameters, and faster inference speed.

IMAGE AND VISION COMPUTING (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

A lightweight network for monocular depth estimation with decoupled body and edge supervision

Usman Ali et al.

Summary: The paper proposes a method to address the inaccuracy of depth prediction near object boundaries by decoupling depth features and explicitly learning body and edge depth maps, improving depth estimation accuracy. Additionally, a lightweight network structure is introduced to efficiently combine multi-scale features.

IMAGE AND VISION COMPUTING (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

MGNet: Monocular Geometric Scene Understanding for Autonomous Driving

Markus Schoen et al.

Summary: MGNet is a multi-task framework for monocular geometric scene understanding, combining panoptic segmentation and self-supervised monocular depth estimation. The model is designed for low latency and real-time inference on a single consumer-grade GPU, producing dense 3D point clouds with instance-aware semantic labels. It demonstrates competitive performance on popular autonomous driving benchmarks Cityscapes and KITTI.

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

Fine-grained Semantics-aware Representation Enhancement for Self-supervised Monocular Depth Estimation

Hyunyoung Jung et al.

Summary: Researchers have proposed a new approach to improve self-supervised monocular depth estimation by leveraging cross-domain information and scene semantics, with two methods developed to enhance geometric representation and feature fusion module. Through evaluation on the KITTI dataset, it has been demonstrated that this method outperforms state-of-the-art methods.

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) (2021)

Add to Collection

Article Engineering, Electrical & Electronic

Unsupervised Person Re-Identification Based on Measurement Axis

Jiahan Li et al.

Summary: The main focus of unsupervised person re-identification is to cluster unlabeled samples in the target domain, with a particular emphasis on mining deep semantic information and improving the combination of source and target domains. By designing a measurement axis component and implementing a new loss function, the study significantly enhances the network's migration ability, resulting in superior accuracy compared to existing techniques in experimental results across two person re-identification domains.

IEEE SIGNAL PROCESSING LETTERS (2021)

Add to Collection

Article Optics