NewEvery arXiv paper, its researchers & institutions — mapped.
the archive

#video analysis

5 results
cs.CV2019

Enhanced 3D convolutional networks for crowd counting

Zhikang Zou, Huiliang Shao, Xiaoye Qu +2

The paper introduces a temporal channel-aware (TCA) block that uses 3D convolutions and channel-wise modulation to capture spatio‑temporal information for more accurate crowd count…

#crowd counting#3d convolution#temporal modeling#video analysis
cs.CV2019

Visual Coin-Tracking: Tracking of Planar Double-Sided Objects

Jonáš Šerých, Jiří Matas

The paper defines the task of tracking rigid planar objects that show both sides (like coins) in video, provides a new benchmark of 17 annotated sequences, and presents a baseline…

#object tracking#planar objects#benchmark dataset#pose estimation
cs.CV2019

Spatially and Temporally Efficient Non-local Attention Network for Video-based Person Re-Identification

Chih-Ting Liu, Chih-Wei Wu, Yu-Chiang Frank Wang +1

The paper introduces a Non-local Video Attention Network (NVAN) and a more efficient variant (STE‑NVAN) that apply non‑local attention at multiple feature levels to improve video‑b…

#person re-identification#video analysis#non-local attention#efficient computation
cs.CV2019

Representation Flow for Action Recognition

AJ Piergiovanni, Michael S. Ryoo

The paper introduces a fully-differentiable convolutional layer that learns optical‑flow‑like motion representations directly within a CNN for action recognition, and shows that st…

#action recognition#optical flow#motion representation#convolutional neural networks
cs.CE2019

Forecasting Short-term Dynamics of Fair-Weather Cumuli using Dynamic Mode Decomposition

Jeff Manning, Ross Baldick

The paper applies Dynamic Mode Decomposition to video sequences of fair‑weather cumulus clouds to forecast short‑term clear‑sky index changes, showing promise for up to seven‑minut…

#dynamic mode decomposition#cloud forecasting#short-term prediction#clear-sky index