paper-with-me

Papers

Tilting at windmills: Data augmentation for deep pose estimation does not help with occlusions

2020-10-20 · Rafal Pytel, Osman Semih Kayhan, Jan C. van Gemert

Occlusion degrades the performance of human pose estimation. In this paper, we introduce targeted keypoint and body part occlusion attacks. The effects of the attacks are systematically analyzed on the best performing methods. In addition, we propose occlusion specific data augmentation techniques against keypoint and part attacks. Our extensive experiments show that human pose estimation methods are not robust to occlusion and data augmentation does not solve the occlusion problems.

📄 PDF Abstract BibTeX arXiv:2010.10451

Code (1)

rpytel1/occlusion-vs-data-augmentations 공식 구현 pytorch

Tasks

Data AugmentationPose Estimation

Similar Papers 제목 키워드 기반

Low-level Pose Control of Tilting Multirotor for Wall Perching Tasks Using Reinforcement Learning

2021-08-11 · Hyungyu Lee, Myeongwoo Jeong, Chanyoung Kim, Hyungtae Lim 외

Recently, needs for unmanned aerial vehicles (UAVs) that are attachable to the wall have been highlighted. As one of the ways to address the need, researches on various tilting multirotors that can increase maneuverabili…

reinforcement-learningReinforcement LearningReinforcement Learning (RL)

Exponential tilting of subweibull distributions

2024-07-16 · F. William Townes

The class of subweibull distributions has recently been shown to generalize the important properties of subexponential and subgaussian random variables. We describe alternative characterizations of subweibull distributio…

Design and Control of a VTOL Aerial Vehicle Tilting its Rotors Only with Rotor Thrusts and a Passive Joint

2023-11-09 · Takumi Ito, Riku Funada, Mitsuji Sampei

This paper presents a novel VTOL UAV that owns a link connecting four rotors and a fuselage by a passive joint, allowing the control of the rotor's tilting angle by adjusting the rotors' thrust. This unique structure con…

Tilting Approximate Models

2018-05-28 · Andreas Tryphonides

Model approximations are common practice when estimating structural or quasi-structural models. The paper considers the econometric properties of estimators that utilize projections to reimpose information about the exac…

counterfactual

Stress-Aware Learning under KL Drift via Trust-Decayed Mirror Descent

2025-10-17 · Gabriel Nixon Raj arxiv

We study sequential decision-making under distribution drift. We propose entropy-regularized trust-decay, which injects stress-aware exponential tilting into both belief updates and mirror-descent decisions. On the simpl…

Distributed Optimization