Search:
Match:
31 results
Paper#SLAM, Computer Vision, Deep Learning🔬 ResearchAnalyzed: Jan 3, 2026 06:15

FoundationSLAM: Dense Visual SLAM with Depth Foundation Models

Published:Dec 31, 2025 17:57
1 min read
ArXiv

Analysis

This paper introduces FoundationSLAM, a novel monocular dense SLAM system that leverages depth foundation models to improve the accuracy and robustness of visual SLAM. The key innovation lies in bridging flow estimation with geometric reasoning, addressing the limitations of previous flow-based approaches. The use of a Hybrid Flow Network, Bi-Consistent Bundle Adjustment Layer, and Reliability-Aware Refinement mechanism are significant contributions towards achieving real-time performance and superior results on challenging datasets. The paper's focus on addressing geometric consistency and achieving real-time performance makes it a valuable contribution to the field.
Reference

“FoundationSLAM achieves superior trajectory accuracy and dense reconstruction quality across multiple challenging datasets, while running in real-time at 18 FPS.”

PermalinkArXiv
Research Paper#Adversarial Attacks, Monocular Depth Estimation, Computer Vision🔬 ResearchAnalyzed: Jan 3, 2026 08:41

Adversarial Attack on Monocular Depth Estimation using Physics-in-the-Loop Optimization

Published:Dec 31, 2025 11:30
1 min read
ArXiv

Analysis

This paper addresses the vulnerability of deep learning models for monocular depth estimation to adversarial attacks. It's significant because it highlights a practical security concern in computer vision applications. The use of Physics-in-the-Loop (PITL) optimization, which considers real-world device specifications and disturbances, adds a layer of realism and practicality to the attack, making the findings more relevant to real-world scenarios. The paper's contribution lies in demonstrating how adversarial examples can be crafted to cause significant depth misestimations, potentially leading to object disappearance in the scene.
Reference

“The proposed method successfully created adversarial examples that lead to depth misestimations, resulting in parts of objects disappearing from the target scene.”

PermalinkArXiv
Research Paper#Robotics, Computer Vision, AI Navigation🔬 ResearchAnalyzed: Jan 3, 2026 15:46

RANGER: Monocular Zero-Shot Semantic Navigation

Published:Dec 30, 2025 13:25
1 min read
ArXiv

Analysis

This paper introduces RANGER, a novel zero-shot semantic navigation framework that addresses limitations of existing methods by operating with a monocular camera and demonstrating strong in-context learning (ICL) capability. It eliminates reliance on depth and pose information, making it suitable for real-world scenarios, and leverages short videos for environment adaptation without fine-tuning. The framework's key components and experimental results highlight its competitive performance and superior ICL adaptability.
Reference

“RANGER achieves competitive performance in terms of navigation success rate and exploration efficiency, while showing superior ICL adaptability.”

PermalinkArXiv
Research Paper#Adversarial Attacks, Monocular Depth Estimation, Autonomous Driving, Diffusion Models🔬 ResearchAnalyzed: Jan 3, 2026 16:49

Adversarial Objects for Depth Estimation Attacks via Diffusion

Published:Dec 30, 2025 09:41
1 min read
ArXiv

Analysis

This paper addresses the vulnerability of monocular depth estimation (MDE) in autonomous driving to adversarial attacks. It proposes a novel method using a diffusion-based generative adversarial attack framework to create realistic and effective adversarial objects. The key innovation lies in generating physically plausible objects that can induce significant depth shifts, overcoming limitations of existing methods in terms of realism, stealthiness, and deployability. This is crucial for improving the robustness and safety of autonomous driving systems.
Reference

“The framework incorporates a Salient Region Selection module and a Jacobian Vector Product Guidance mechanism to generate physically plausible adversarial objects.”

PermalinkArXiv
Research Paper#Computer Vision, Medical Robotics, Depth Estimation🔬 ResearchAnalyzed: Jan 3, 2026 16:02

Improving Depth Estimation in Robotic Surgery with Synthetic Data

Published:Dec 29, 2025 17:29
1 min read
ArXiv

Analysis

This paper addresses a critical challenge in robotic surgery: accurate depth estimation in challenging environments. It leverages synthetic data and a novel adaptation technique (DV-LORA) to improve performance, particularly in the presence of specular reflections and transparent surfaces. The introduction of a new evaluation protocol is also significant. The results demonstrate a substantial improvement over existing methods, making this work valuable for the field.
Reference

“Achieving an accuracy (< 1.25) of 98.1% and reducing Squared Relative Error by over 17% compared to established baselines.”

PermalinkArXiv
Research Paper#Computer Vision, Depth Estimation, Generative Models🔬 ResearchAnalyzed: Jan 3, 2026 19:47

Visual Autoregressive Depth Estimation

Published:Dec 27, 2025 17:08
1 min read
ArXiv

Analysis

This paper introduces a novel approach to monocular depth estimation using visual autoregressive (VAR) priors, offering an alternative to diffusion-based methods. It leverages a text-to-image VAR model and introduces a scale-wise conditional upsampling mechanism. The method's efficiency, requiring only 74K synthetic samples for fine-tuning, and its strong performance, particularly in indoor benchmarks, are noteworthy. The work positions autoregressive priors as a viable generative model family for depth estimation, emphasizing data scalability and adaptability to 3D vision tasks.
Reference

“The method achieves state-of-the-art performance in indoor benchmarks under constrained training conditions.”

PermalinkArXiv
Research Paper#Computer Vision, Pose Estimation, Transformers🔬 ResearchAnalyzed: Jan 3, 2026 16:24

KV-Tracker: Real-Time Pose Tracking with Transformers

Published:Dec 27, 2025 13:02
1 min read
ArXiv

Analysis

This paper addresses the computational bottleneck of multi-view 3D geometry networks for real-time applications. It introduces KV-Tracker, a novel method that leverages key-value (KV) caching within a Transformer architecture to achieve significant speedups in 6-DoF pose tracking and online reconstruction from monocular RGB videos. The model-agnostic nature of the caching strategy is a key advantage, allowing for application to existing multi-view networks without retraining. The paper's focus on real-time performance and the ability to handle challenging tasks like object tracking and reconstruction without depth measurements or object priors are significant contributions.
Reference

“The caching strategy is model-agnostic and can be applied to other off-the-shelf multi-view networks without retraining.”

PermalinkArXiv
Research Paper#3D Reconstruction, Remote Sensing, Foundation Models, Urban Modeling🔬 ResearchAnalyzed: Jan 3, 2026 16:28

SAM 3D for 3D Building Reconstruction from Remote Sensing Images

Published:Dec 27, 2025 03:47
1 min read
ArXiv

Analysis

This paper introduces and evaluates the use of SAM 3D, a general-purpose image-to-3D foundation model, for monocular 3D building reconstruction from remote sensing imagery. It's significant because it explores the application of a foundation model to a specific domain (urban modeling) and provides a benchmark against an existing method (TRELLIS). The paper highlights the potential of foundation models in this area and identifies limitations and future research directions, offering practical guidance for researchers.
Reference

“SAM 3D produces more coherent roof geometry and sharper boundaries compared to TRELLIS.”

PermalinkArXiv
Research#Drone Racing🔬 ResearchAnalyzed: Jan 10, 2026 08:02

Advanced Drone Racing: Combining VIO and Perception for Autonomous Flight

Published:Dec 23, 2025 16:12
1 min read
ArXiv

Analysis

This research explores a crucial area for autonomous drone applications, specifically within the demanding environment of drone racing. The use of drift-corrected monocular VIO and perception-aware planning signifies a step forward in real-time control and adaptability.
Reference

“The research focuses on drift-corrected monocular VIO and perception-aware planning.”

PermalinkArXiv
Research#3D Vision🔬 ResearchAnalyzed: Jan 10, 2026 08:51

VOIC: Advancing 3D Scene Understanding from Single Images

Published:Dec 22, 2025 02:05
1 min read
ArXiv

Analysis

The research paper on VOIC introduces a novel approach to monocular 3D semantic scene completion, potentially improving the accuracy of environmental perception. This method could be significant for applications like autonomous driving and robotics, which require a detailed understanding of their surroundings.
Reference

“The research is published on ArXiv.”

PermalinkArXiv
Research#3D Reconstruction🔬 ResearchAnalyzed: Jan 10, 2026 09:16

Novel Approach to Large-Scale 3D Reconstruction from Monocular Images

Published:Dec 20, 2025 06:37
1 min read
ArXiv

Analysis

This research explores a new method for 3D reconstruction using a single camera, addressing the challenges of large-scale environments. The joint learning approach, incorporating depth, pose, and local radiance fields, is a promising step in improving reconstruction accuracy and efficiency.
Reference

“The research focuses on using a single camera (monocular) for 3D reconstruction.”

PermalinkArXiv
Research#Depth Estimation🔬 ResearchAnalyzed: Jan 10, 2026 09:18

EndoStreamDepth: Advancing Monocular Depth Estimation for Endoscopic Videos

Published:Dec 20, 2025 00:53
1 min read
ArXiv

Analysis

This research, published on ArXiv, focuses on temporal consistency in monocular depth estimation for endoscopic videos. The advancements in this area have the potential to significantly improve surgical procedures and diagnostics.
Reference

“The research focuses on temporally consistent monocular depth estimation.”

PermalinkArXiv
Research#Robotics🔬 ResearchAnalyzed: Jan 10, 2026 09:21

SurgiPose: Advancing Surgical Robotics with Monocular Video Kinematics

Published:Dec 19, 2025 21:15
1 min read
ArXiv

Analysis

The SurgiPose project, detailed on ArXiv, represents a significant step towards enabling more sophisticated surgical robot learning. The method's reliance on monocular video offers a potentially more accessible and cost-effective approach compared to methods requiring stereo vision or other specialized sensors.
Reference

“The paper focuses on estimating surgical tool kinematics from monocular video for surgical robot learning.”

PermalinkArXiv
Research#robotics🔬 ResearchAnalyzed: Jan 4, 2026 07:15

Flying in Clutter on Monocular RGB by Learning in 3D Radiance Fields with Domain Adaptation

Published:Dec 19, 2025 08:44
1 min read
ArXiv

Analysis

This article describes a research paper focusing on a specific problem in computer vision and robotics: enabling autonomous navigation in complex, cluttered environments using only monocular RGB images. The approach involves learning 3D representations (radiance fields) and adapting them to different visual domains. The title suggests a focus on practical application (flying) and the challenges of real-world environments (clutter). The use of 'domain adaptation' indicates an attempt to generalize the learned models across different visual conditions.
Reference

“”

PermalinkArXiv
Research#llm🔬 ResearchAnalyzed: Jan 4, 2026 09:29

PoseMoE: Mixture-of-Experts Network for Monocular 3D Human Pose Estimation

Published:Dec 18, 2025 13:01
1 min read
ArXiv

Analysis

The article introduces PoseMoE, a novel approach using a Mixture-of-Experts (MoE) network for 3D human pose estimation from monocular images. This suggests an advancement in the field by potentially improving accuracy and efficiency compared to existing methods. The use of MoE implies the model can handle complex data variations and learn specialized representations.
Reference

“N/A - This is an abstract, not a news article with quotes.”

PermalinkArXiv
Research#llm🔬 ResearchAnalyzed: Jan 4, 2026 08:32

MoonSeg3R: Monocular Online Zero-Shot Segment Anything in 3D with Reconstructive Foundation Priors

Published:Dec 17, 2025 16:36
1 min read
ArXiv

Analysis

This article introduces MoonSeg3R, a novel approach for 3D segmentation. The core innovation lies in its ability to perform zero-shot segmentation, meaning it can segment objects without prior training on specific object classes. It leverages reconstructive foundation priors, suggesting a focus on learning from underlying data structures to improve segmentation accuracy and efficiency. The 'monocular online' aspect implies the system operates using a single camera and processes data in real-time.
Reference

“The article is based on a paper from ArXiv, suggesting it's a research paper.”

PermalinkArXiv
Research#Scene Simulation🔬 ResearchAnalyzed: Jan 10, 2026 10:39

CRISP: Advancing Real-World Scene Simulation from Single-View Video

Published:Dec 16, 2025 18:59
1 min read
ArXiv

Analysis

This research explores a novel method for creating realistic simulations from monocular videos, a crucial area for robotics and virtual reality. The paper's focus on contact-guided simulation using planar scene primitives suggests a promising avenue for improved scene understanding and realistic interactions.
Reference

“The research originates from ArXiv, a platform for pre-print scientific papers.”

PermalinkArXiv
Research#Depth Estimation🔬 ResearchAnalyzed: Jan 10, 2026 10:43

DASP: Improving Nighttime Depth Estimation Through Self-Supervised Learning and Domain Adaptation

Published:Dec 16, 2025 16:11
1 min read
ArXiv

Analysis

This research paper presents a novel approach to address a challenging computer vision problem: monocular depth estimation in nighttime environments. The use of self-supervised learning and domain adaptation techniques suggests a robust methodology for improving performance in low-light conditions.
Reference

“The paper focuses on self-supervised nighttime monocular depth estimation.”

PermalinkArXiv
Research#View Synthesis🔬 ResearchAnalyzed: Jan 10, 2026 10:46

Expanding Dynamic Scene View Synthesis from Single-Camera Footage

Published:Dec 16, 2025 13:43
1 min read
ArXiv

Analysis

This research explores a novel approach to create 3D views from monocular videos of dynamic scenes. The constrained nature of the input data presents a significant challenge, making this a noteworthy contribution to computer vision.
Reference

“The research focuses on view synthesis.”

PermalinkArXiv
Research#Depth Completion🔬 ResearchAnalyzed: Jan 10, 2026 11:12

StarryGazer: Advancing Depth Image Completion with Domain-Agnostic AI

Published:Dec 15, 2025 09:56
1 min read
ArXiv

Analysis

This ArXiv paper explores a novel approach to completing single depth images, a challenging task in computer vision. The domain-agnostic nature of the model suggests potential for broad applicability across different scenarios and datasets.
Reference

“The research focuses on leveraging Monocular Depth Estimation models.”

PermalinkArXiv
Research#computer vision🔬 ResearchAnalyzed: Jan 4, 2026 09:10

BokehDepth: Enhancing Monocular Depth Estimation through Bokeh Generation

Published:Dec 13, 2025 18:39
1 min read
ArXiv

Analysis

This article introduces BokehDepth, a method for improving monocular depth estimation. The core idea is to leverage bokeh generation, likely to provide additional visual cues for depth perception. The source being ArXiv suggests this is a research paper, and the focus is on a specific technical approach within the field of computer vision.

Key Takeaways

    Reference

    “”

    PermalinkArXiv
    Research#Motion Capture🔬 ResearchAnalyzed: Jan 10, 2026 11:57

    MoCapAnything: Revolutionizing 3D Motion Capture from Single-View Videos

    Published:Dec 11, 2025 18:09
    1 min read
    ArXiv

    Analysis

    The research paper on MoCapAnything introduces a potentially significant advancement in 3D motion capture technology, enabling the capture of arbitrary skeletons from monocular videos. This could have a broad impact on various fields, from animation and gaming to robotics and human-computer interaction.
    Reference

    “The technology captures 3D motion from single-view (monocular) videos.”

    PermalinkArXiv
    Research#llm🔬 ResearchAnalyzed: Jan 4, 2026 08:12

    Sharp Monocular View Synthesis in Less Than a Second

    Published:Dec 11, 2025 14:34
    1 min read
    ArXiv

    Analysis

    The article title suggests a significant advancement in computer vision, specifically in the area of view synthesis. The claim of speed (less than a second) is a key selling point, implying efficiency. The use of 'monocular' indicates the system works from a single image, which is a common challenge in this field. The source, ArXiv, suggests this is a research paper, likely detailing a new algorithm or technique.
    Reference

    “”

    PermalinkArXiv
    Research#video processing🔬 ResearchAnalyzed: Jan 4, 2026 10:31

    StereoWorld: Geometry-Aware Monocular-to-Stereo Video Generation

    Published:Dec 10, 2025 06:50
    1 min read
    ArXiv

    Analysis

    The article introduces a research paper on generating stereo video from monocular video, focusing on geometric understanding. This suggests advancements in video processing and potentially in applications like VR/AR content creation. The 'geometry-aware' aspect is key, implying the use of depth estimation or 3D reconstruction techniques. The source being ArXiv indicates this is a preliminary research finding, not yet peer-reviewed.

    Key Takeaways

      Reference

      “”

      PermalinkArXiv
      Research#computer vision🔬 ResearchAnalyzed: Jan 4, 2026 09:23

      Relightable and Dynamic Gaussian Avatar Reconstruction from Monocular Video

      Published:Dec 10, 2025 05:51
      1 min read
      ArXiv

      Analysis

      This article describes a research paper on reconstructing avatars from a single video source. The focus is on creating avatars that can be relit and are dynamic, using Gaussian splatting techniques. The source is ArXiv, indicating it's a pre-print and likely targets a technical audience. The core innovation likely lies in the method of representing the avatar (Gaussian splatting) and its ability to handle relighting and dynamic movement.
      Reference

      “”

      PermalinkArXiv
      Research#Depth Estimation🔬 ResearchAnalyzed: Jan 10, 2026 12:33

      Advanced Monocular Depth Estimation Using Scale-Invariant and View-Relational Learning

      Published:Dec 9, 2025 15:17
      1 min read
      ArXiv

      Analysis

      This research explores a novel approach to monocular depth estimation, a crucial task in computer vision. The study's focus on scale-invariance and view-relational learning suggests advancements in handling complex scenes and improving depth accuracy from a single camera.
      Reference

      “The research focuses on full surround monocular depth.”

      PermalinkArXiv
      Research#SLAM🔬 ResearchAnalyzed: Jan 10, 2026 12:34

      OpenMonoGS-SLAM: Advancing Monocular SLAM with Gaussian Splatting and Open-Set Semantics

      Published:Dec 9, 2025 14:10
      1 min read
      ArXiv

      Analysis

      This research introduces a novel approach to monocular SLAM using Gaussian Splatting and open-set semantics, likely improving scene understanding. The paper's focus on open-set semantics suggests an attempt to handle unknown objects more effectively within SLAM environments.
      Reference

      “The research is published on ArXiv.”

      PermalinkArXiv
      Research#3D Tracking🔬 ResearchAnalyzed: Jan 10, 2026 12:38

      TrackingWorld: Pioneering World-Centric 3D Tracking with a Single Camera

      Published:Dec 9, 2025 08:35
      1 min read
      ArXiv

      Analysis

      This research from ArXiv presents a novel approach to 3D object tracking, utilizing a single camera to achieve world-centric tracking of most pixels. The paper's focus on monocular vision and comprehensive pixel tracking suggests a potential breakthrough in areas like robotics and autonomous systems.
      Reference

      “TrackingWorld focuses on world-centric monocular 3D tracking.”

      PermalinkArXiv
      Research#Depth Estimation🔬 ResearchAnalyzed: Jan 10, 2026 12:41

      Beyond Accuracy: Human-Likeness in Monocular Depth Estimation Remains Elusive

      Published:Dec 9, 2025 01:42
      1 min read
      ArXiv

      Analysis

      This ArXiv paper highlights a critical distinction in monocular depth estimation, emphasizing that achieving high accuracy doesn't automatically equate to human-like understanding of scene depth. It encourages researchers to focus on developing models that capture the nuances of human visual perception beyond simple numerical precision.
      Reference

      “The paper focuses on monocular depth estimation, using only a single camera to estimate the depth of a scene.”

      PermalinkArXiv
      Research#3D Detection🔬 ResearchAnalyzed: Jan 10, 2026 13:02

      LeAD-M3D: Enhancing Real-time 3D Object Detection with Asymmetric Distillation

      Published:Dec 5, 2025 12:08
      1 min read
      ArXiv

      Analysis

      The paper presents LeAD-M3D, a novel approach for real-time monocular 3D detection using asymmetric distillation. This research contributes to the field by improving the accuracy and efficiency of 3D object detection from a single camera view.
      Reference

      “The research is sourced from ArXiv.”

      PermalinkArXiv
      Research#llm🔬 ResearchAnalyzed: Jan 4, 2026 09:32

      MDE-AgriVLN: Agricultural Vision-and-Language Navigation with Monocular Depth Estimation

      Published:Dec 3, 2025 16:52
      1 min read
      ArXiv

      Analysis

      This article introduces a research paper on agricultural navigation using vision and language models, incorporating monocular depth estimation. The focus is on applying AI to agricultural tasks, specifically navigation. The use of monocular depth estimation suggests an attempt to improve the accuracy and robustness of the navigation system in complex agricultural environments. The source being ArXiv indicates this is a preliminary research paper, not yet peer-reviewed.
      Reference

      “”

      PermalinkArXiv