Filters








104,710 Hits in 6.7 sec

Object Detection in Videos by High Quality Object Linking [article]

Peng Tang, Chunyu Wang, Xinggang Wang, Wenyu Liu, Wenjun Zeng, Jingdong Wang
2019 arXiv   pre-print
detection network that detects short tubelets in short video segments; (3) a short tubelet linking algorithm that links temporally-overlapping short tubelets to form long tubelets.  ...  Compared with object detection in static images, object detection in videos is more challenging due to degraded image qualities.  ...  When linking objects over the whole video to consider long range temporal context, there is significant improvements (5.4% to static and 4.0% to without short tubelet linking).  ... 
arXiv:1801.09823v3 fatcat:xabl5dyg2jbu5bqbiww6jpfkxy

Object Detection in Videos by High Quality Object Linking

Peng Tang, Chunyu Wang, Xinggang Wang, Wenyu Liu, Wenjun Zeng, Jingdong Wang
2019 IEEE Transactions on Pattern Analysis and Machine Intelligence  
detection network that detects short tubelets in short video segments; (3) a short tubelet linking algorithm that links temporally-overlapping short tubelets to form long tubelets.  ...  Compared with object detection in static images, object detection in videos is more challenging due to degraded image qualities.  ...  This problem does not occur in the short tubelet linking stage because we allow broken links for long range linking.  ... 
doi:10.1109/tpami.2019.2910529 pmid:30990176 fatcat:o7vskzkxzjhktayigtraknq2ja

Long Short-Term Relation Networks for Video Action Detection [article]

Dong Li and Ting Yao and Zhaofan Qiu and Houqiang Li and Tao Mei
2020 arXiv   pre-print
This motivates us to capture both short-term and long-term relations in a video.  ...  a long-range span of the video.  ...  By consolidating the idea of modeling both short-term and longterm relation in a video, we novelly present Long Short-Term Relation Networks (LSTR) for boosting video action detection.  ... 
arXiv:2003.14065v1 fatcat:lam2eqnyrjadzgibz4vlcpmyc4

MeMOT: Multi-Object Tracking with Memory [article]

Jiarui Cai, Mingze Xu, Wei Li, Yuanjun Xiong, Wei Xia, Zhuowen Tu, Stefano Soatto
2022 arXiv   pre-print
We propose an online tracking algorithm that performs the object detection and data association under a common framework, capable of linking objects after a long time span.  ...  This is realized by preserving a large spatio-temporal memory to store the identity embeddings of the tracked objects, and by adaptively referencing and aggregating useful information from the memory as  ...  stage that links the detected object instances across time [5, 70] by modeling the state changes of tracked objects and solving a matching problem between them and the detection results.  ... 
arXiv:2203.16761v1 fatcat:6arhpggwivgybdnhehyj2jm7se

Spatio-temporal Tubelet Feature Aggregation and Object Linking in Videos [article]

Daniel Cores, Víctor M. Brea, Manuel Mucientes
2020 arXiv   pre-print
We propose a two stage object detector called FANet based on short-term spatio-temporal feature aggregation to give a first detection set, and long-term object linking to refine these detections.  ...  Finally, a long-term linking method builds long tubes using the previously calculated short tubelets to overcome detection errors.  ...  ACKNOWLEDGMENT This research was partially funded by the Spanish Ministry of Science, Innovation and Universities under grants TIN2017-84796-C2-1-R and RTI2018-097088-B-C32, and the Galician Ministry of  ... 
arXiv:2004.00451v2 fatcat:oxflf7wvubbzjda23gffmd7u5u

Towards Longer Long-Range Motion Trajectories

Michael Rubinstein, Ce Liu
2012 Procedings of the British Machine Vision Conference 2012  
Our algorithm re-correlates the short trajectories and links them to form a long-range motion representation by formulating a combinatorial assignment problem that is defined and optimized globally over  ...  We leverage accurate local (short-range) trajectories produced by current motion tracking methods and use them as an initial estimate for a global (long-range) solution.  ...  This material is based upon work supported by the National Science Foundation under Grant No. CGV 1111415 and by an NVDIA Fellowship to M. Rubinstein.  ... 
doi:10.5244/c.26.53 dblp:conf/bmvc/RubinsteinL12 fatcat:xkcmywfiabbitdcs4harzacbee

A testing framework for background subtraction algorithms comparison in intrusion detection context

Corentin Lallier, Emanuelle Reynaud, Lionel Robinault, Laure Tougne
2011 2011 8th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)  
Identifying objects from a video stream is a fundamental and critical task in many computer-vision applications.  ...  We focus here on a particular application of computer vision: intrusion detection in video surveillance.  ...  In the object extraction process, short or long range errors are less important than medium range error, because pixels on medium range impact greatly on object shapes recognition.  ... 
doi:10.1109/avss.2011.6027343 dblp:conf/avss/LallierRRT11 fatcat:sezvhwmf25bvhoswhy7t3elwgu

Audio-visual atoms for generic video concept classification

Wei Jiang, Courtenay Cotton, Shih-Fu Chang, Dan Ellis, Alexander C. Loui
2010 ACM Transactions on Multimedia Computing, Communications, and Applications (TOMCCAP)  
We develop a hierarchical algorithm to extract visual atoms from generic videos, and locate energy onsets from the corresponding soundtrack by time-frequency analysis.  ...  We investigate the challenging issue of joint audio-visual analysis of generic videos targeting at concept detection.  ...  Long-Term Linking to Generate Visual Atoms By now, for an input video segment u, over each short-term video slice v k , k = 1, . . . , K, we have a set of N k r short-term region tracks r k 1 , . . . ,  ... 
doi:10.1145/1823746.1823748 fatcat:ayq6ce4wsvhg5i3o25uqc2ckoy

Social Fabric: Tubelet Compositions for Video Relation Detection [article]

Shuo Chen, Zenglin Shi, Pascal Mettes, Cees G. M. Snoek
2021 arXiv   pre-print
This paper strives to classify and detect the relationship between object tubelets appearing within a video as a triplet.  ...  These primitives are learned over all relations, resulting in a compact representation able to localize and classify relations from the pool of co-occurring object tubelets across all timespans in a video  ...  The authors thank Pengwan Yang for his help on figure design and comments.  ... 
arXiv:2108.08363v1 fatcat:okjgeetdr5adrnut4hrrwcadki

Perception Principles Guided Video Segmentation

Cheng Chen, Guoliang Fan
2005 2005 IEEE 7th Workshop on Multimedia Signal Processing  
The proposed algorithm employs both long-range motion information, i.e., trajectory, and short-range motion information, i.e., change detection, to retain temporal continuity and spatial homogeneity of  ...  In the third layer, still and moving regions are merged into background and moving objects by a graph-based approach with different similarity metrics.  ...  combines both short-range and long-range motion information.  ... 
doi:10.1109/mmsp.2005.248664 dblp:conf/mmsp/ChenF05 fatcat:yg5643wc2fhavkegcnqz7lrw4q

Towards Long-Form Video Understanding [article]

Chao-Yuan Wu, Philipp Krähenbühl
2021 arXiv   pre-print
We introduce a framework for modeling long-form videos and develop evaluation protocols on large-scale datasets.  ...  These systems understand the present, but fail to contextualize it in past or future events. In this paper, we study long-form video understanding.  ...  This material is based upon work supported by the National Science Foundation under Grant No. IIS-1845485 and IIS-2006820, and the NSF Institute for Foundations of Machine Learning.  ... 
arXiv:2106.11310v1 fatcat:vxw7ugggwbeibd43ubkaavuw3y

Plug Play Convolutional Regression Tracker for Video Object Detection [article]

Ye Lyu, Michael Ying Yang, George Vosselman, Gui-Song Xia
2020 arXiv   pre-print
Video object detection targets to simultaneously localize the bounding boxes of the objects and identify their classes in a given video.  ...  In this paper, we propose a Plug & Play scale-adaptive convolutional regression tracker for the video object detection task, which could be easily and compatibly implanted into the current state-of-the-art  ...  The problem for short term feature aggregation [43, 1, 47, 41] is that they pre-define a limited range of frames for the detection in each frame and the long range coherency cannot be preserved and utilized  ... 
arXiv:2003.00981v1 fatcat:qkjzxuwjkjbuzkf2jfr25u2jgm

SMOT: Single-Shot Multi Object Tracking [article]

Wei Li, Yuanjun Xiong, Shuo Yang, Siqi Deng, Wei Xia
2020 arXiv   pre-print
On three benchmarks of object tracking: Hannah, Music Videos, and MOT17, the proposed SMOT achieves state-of-the-art performance.  ...  Contrary to the existing tracking by detection approaches which suffer from errors made by the object detectors, SMOT adopts the recently proposed scheme of tracking by re-detection.  ...  Then the third stage, long-term tracklets linking, links the tracklets in the whole video to form long-term tracking outputs.  ... 
arXiv:2010.16031v1 fatcat:stvpohnb6vehfgehlpjhkuvybi

Implicit Motion Handling for Video Camouflaged Object Detection [article]

Xuelian Cheng, Huan Xiong, Deng-Ping Fan, Yiran Zhong, Mehrtash Harandi, Tom Drummond, Zongyuan Ge
2022 arXiv   pre-print
We propose a new video camouflaged object detection (VCOD) framework that can exploit both short-term dynamics and long-term temporal consistency to detect camouflaged objects from video frames.  ...  Therefore, effectively handling temporal dynamics in videos becomes the key for the VCOD task as the camouflaged objects will be noticeable when they move.  ...  Abstract We propose a new video camouflaged object detection (VCOD) framework that can exploit both short-term dynamics and long-term temporal consistency to detect camouflaged objects from video frames  ... 
arXiv:2203.07363v2 fatcat:wnba2jnwjndx5ddtcqpkeijntq

Globally optimal solution to multi-object tracking with merged measurements

Joao F. Henriques, Rui Caseiro, Jorge Batista
2011 2011 International Conference on Computer Vision  
Multiple object tracking has been formulated recently as a global optimization problem, and solved efficiently with optimal methods such as the Hungarian Algorithm.  ...  A severe limitation is the inability to model multiple objects that are merged into a single measurement, and track them as a group, while retaining optimality.  ...  Objects in a group are identified by nested outlines with their respective colors. Trails show the short-term trajectory of each object.  ... 
doi:10.1109/iccv.2011.6126532 dblp:conf/iccv/HenriquesCB11 fatcat:3azlb3gvsbblbg46ivsgri2v4i
« Previous Showing results 1 — 15 out of 104,710 results