Filters








1,621 Hits in 3.1 sec

Towards Automatic Model Specialization for Edge Video Analytics [article]

Daniel Rivas, Francesc Guim, Jordà Polo, Pubudu M. Silva, Josep Ll. Berral, David Carrera
2021 arXiv   pre-print
Yet, there is a need for a meeting point between the edge and accurate real-time video analytics.  ...  In this paper, we present and evaluate COVA (Contextually Optimized Video Analytics), a framework to assist in the automatic specialization of models for video analytics in edge cameras.  ...  In closing, we consider COVA to be the first step towards fully automated large-scale deployments for real-time edge video analytics.  ... 
arXiv:2104.06826v2 fatcat:uoirbiby7fem7brgnub2yjhuhe

Deployment of Customized Deep Learning based Video Analytics On Surveillance Cameras [article]

Pratik Dubal, Rohan Mahadev, Suraj Kothawade, Kunal Dargan, Rishabh Iyer
2018 arXiv   pre-print
Towards this end, we describe our data processing and model training pipeline, which can train and fine-tune models from videos with a quick turnaround time.  ...  By sharing our implementation details and the experiences learned from deploying customized deep learning models for various customers, we hope that customized deep learning based video analytics is widely  ...  Towards this end, we perform transfer learning and fine-tuning on the dataset consisting of images from the deployment scenario.  ... 
arXiv:1805.10604v2 fatcat:xzckmwvyv5hvfik35awnvwc2ji

SMURF: Self-Teaching Multi-Frame Unsupervised RAFT with Full-Image Warping [article]

Austin Stone, Daniel Maurer, Alper Ayvaci, Anelia Angelova, Rico Jonschkowski
2021 arXiv   pre-print
for handling out-of-frame motion, and an approach for learning effectively from multi-frame video data while still only requiring two frames for inference.  ...  We present SMURF, a method for unsupervised learning of optical flow that improves state of the art on all benchmarks by 36% to 40% (over the prior best method UFlow) and even outperforms several supervised  ...  Without fine-tuning, our flow model estimates stereo depth "zero-shot" at an accuracy comparable to state of the art unsupervised methods trained for that task.  ... 
arXiv:2105.07014v1 fatcat:qfjvti4avjafdbzoodcl2aohoy

Hierarchical Relational Networks for Group Activity Recognition and Retrieval [chapter]

Mostafa S. Ibrahim, Greg Mori
2018 Lecture Notes in Computer Science  
Second, we propose a Relational Autoencoder model for unsupervised learning of features for action and scene retrieval.  ...  First, given a video sequence of people doing a collective activity, the relational scene representation is utilized for multi-person activity recognition.  ...  Group activity recognition arises in the context of multi-person scenes, including in video surveillance, sports analytics, and video search and retrieval.  ... 
doi:10.1007/978-3-030-01219-9_44 fatcat:hwccyxsaejdqpgainxu7u7pbnm

Deep learning applications and challenges in big data analytics

Maryam M Najafabadi, Flavio Villanustre, Taghi M Khoshgoftaar, Naeem Seliya, Randall Wald, Edin Muharemagic
2015 Journal of Big Data  
A key benefit of Deep Learning is the analysis and learning of massive amounts of unsupervised data, making it a valuable tool for Big Data Analytics where raw data is largely unlabeled and un-categorized  ...  In the present study, we explore how Deep Learning can be utilized for addressing some important problems in Big Data Analytics, including extracting complex patterns from massive volumes of data, semantic  ...  At the end the whole network is fine-tuned by providing supervised data to it.  ... 
doi:10.1186/s40537-014-0007-7 fatcat:65mi6dnv5rg6poesotupqbsm7y

Unsupervised Learning of Depth and Ego-Motion from Monocular Video Using 3D Geometric Constraints [article]

Reza Mahjourian, Martin Wicke, Anelia Angelova
2018 arXiv   pre-print
We present a novel approach for unsupervised learning of depth and ego-motion from monocular video.  ...  Unsupervised learning removes the need for separate supervisory signals (depth or ego-motion ground truth, or multi-view video).  ...  videos.  ... 
arXiv:1802.05522v2 fatcat:vq7w353yvbeibbmnmrfhv457mi

Self-Building Artificial Intelligence and Machine Learning to Empower Big Data Analytics in Smart Cities

Damminda Alahakoon, Rashmika Nawaratne, Yan Xu, Daswin De Silva, Uthayasankar Sivarajah, Bhumika Gupta
2020 Information Systems Frontiers  
This study explores the benefits of self-building AI and machine learning with unsupervised learning for empowering big data analytics for smart city environments.  ...  The new paradigm of the self-building AI and its value are demonstrated using the IoT, video surveillance, and action recognition applications.  ...  Acknowledgements This work was supported by the Data to Decisions Cooperative Research Centre (D2D CRC) as part of their analytics and decision support program and a La Trobe University Postgraduate Research  ... 
doi:10.1007/s10796-020-10056-x fatcat:r5l5roo4ubc45b5paxak7ftkwa

Unsupervised Learning of Depth and Ego-Motion from Monocular Video Using 3D Geometric Constraints

Reza Mahjourian, Martin Wicke, Anelia Angelova
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
We present a novel approach for unsupervised learning of depth and ego-motion from monocular video.  ...  Unsupervised learning removes the need for separate supervisory signals (depth or ego-motion ground truth, or multi-view video).  ...  videos.  ... 
doi:10.1109/cvpr.2018.00594 dblp:conf/cvpr/MahjourianWA18 fatcat:vvjhfpn4qvdkffujq747pxfk4e

The Hessian Penalty: A Weak Prior for Unsupervised Disentanglement [article]

William Peebles, John Peebles, Jun-Yan Zhu, Alexei Efros, Antonio Torralba
2020 arXiv   pre-print
Existing disentanglement methods for deep generative models rely on hand-picked priors and complex encoder-based architectures.  ...  Additionally, we use our regularization term to identify interpretable directions in BigGAN's latent space in an unsupervised fashion.  ...  We thank Pieter Abbeel, Taesung Park, Richard Zhang, Mathieu Aubry, Ilija Radosavovic, Tim Brooks, Karttikeya Mangalam, and all of BAIR for valuable discussions and encouragement.  ... 
arXiv:2008.10599v1 fatcat:odaoj42xjzggrpstpzn4xr2n4a

EIQIS: Toward an Event-Oriented Indexable and Queryable Intelligent Surveillance System [article]

Seyed Yahya Nikouei, Yu Chen, Alexander Aved, Erik Blasch
2018 arXiv   pre-print
A measure of intelligence is not what is known, but is recalled, hence, future edge intelligence must provide recalled information for dynamic response.  ...  Edge computing provides the ability to link distributor users for multimedia content, while retaining the power of significant data storage and access at a centralized computer.  ...  The decisions making and the detection of false alarm and the features that raised the alarm are sent for future fine tuning of the algorithms and also some analytical studies.  ... 
arXiv:1807.11329v1 fatcat:vuszt24ly5eyzl4qxztmd2wr44

The Multi-Agent Behavior Dataset: Mouse Dyadic Social Interactions [article]

Jennifer J. Sun, Tomomi Karigo, Dipam Chakraborty, Sharada P. Mohanty, Benjamin Wild, Quan Sun, Chen Chen, David J. Anderson, Pietro Perona, Yisong Yue, Ann Kennedy
2021 arXiv   pre-print
Our dataset consists of trajectory data of social interactions, recorded from videos of freely behaving mice in a standard resident-intruder assay.  ...  datasets all annotated by a single annotator, (2) for style transfer to learn inter-annotator differences in behavior definitions, and (3) for learning of new behaviors of interest given limited training  ...  We are grateful to the team at AICrowd for the support and hosting of our dataset challenge, as well as Northwestern University and Amazon Sagemaker for funding our challenge prizes.  ... 
arXiv:2104.02710v4 fatcat:rgdd7xqouzbi7j5jvimmngwjr4

Semi-supervised Transfer Learning for Image Rain Removal [article]

Wei Wei, Deyu Meng, Qian Zhao, Zongben Xu, Ying Wu
2019 arXiv   pre-print
However, previous deep learning methods need to pre-collect a large set of image pairs with/without synthesized rain for training, which tends to make the neural network be biased toward learning the specific  ...  To this issue, this paper firstly proposes a semi-supervised learning paradigm toward this task.  ...  By utilizing the above encoding manner, we can also construct an objective function for unsupervised rainy images, which can be further used to fine-tune the network parameters through back-propagating  ... 
arXiv:1807.11078v2 fatcat:oj7s62tcifbqpppw7gd4wrhkf4

Semi-Supervised Transfer Learning for Image Rain Removal

Wei Wei, Deyu Meng, Qian Zhao, Zongben Xu, Ying Wu
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
However, previous deep learning methods need to pre-collect a large set of image pairs with/without synthesized rain for training, which tends to make the neural network be biased toward learning the specific  ...  To this issue, this paper firstly proposes a semi-supervised learning paradigm toward this task.  ...  By utilizing the above encoding manner, we can also construct an objective function for unsupervised rainy images, which can be further used to fine-tune the network parameters through back-propagating  ... 
doi:10.1109/cvpr.2019.00400 dblp:conf/cvpr/WeiMZXW19 fatcat:z3ociby7ofed5f5pujz5yo2j34

Federated Self-Supervised Learning of Multi-Sensor Representations for Embedded Intelligence

Aaqib Saeed, Flora D. Salim, Tanir Ozcelebi, Johan Lukkien
2020 IEEE Internet of Things Journal  
Smartphones, wearables, and Internet of Things (IoT) devices produce a wealth of data that cannot be accumulated in a centralized repository for learning supervised models due to privacy, bandwidth limitations  ...  Federated learning provides a compelling framework for learning models from decentralized data, but conventionally, it assumes the availability of labeled samples, whereas on-device data are generally  ...  Consequently, the unsupervised model can be used as a semantic feature extractor or initialization for efficiently adapting to an end-task of interest through fine-tuning with few-labeled instances.  ... 
doi:10.1109/jiot.2020.3009358 fatcat:ylwl4dvr2rczdlxar77d7bcxkq

Discovery of Visual Semantics by Unsupervised and Self-Supervised Representation Learning [article]

Gustav Larsson
2017 arXiv   pre-print
We show that traditional methods for unsupervised learning, such as layer-wise clustering or autoencoders, remain inferior to supervised pre-training.  ...  To address this concern, with the long-term goal of leveraging the abundance of cheap unlabeled data, we explore methods of unsupervised "pre-training."  ...  Next, we look at how much fine-tuning is required for the downstream task.  ... 
arXiv:1708.05812v1 fatcat:w77w3q3ms5c5fnyzl65mkj4ozy
« Previous Showing results 1 — 15 out of 1,621 results