Human Action Recognition from RGB-D Frames Based on Real-Time 3D Optical Flow Estimation

Ballin, Gioia; Munaro, Matteo; Menegatti, Emanuele

doi:10.1007/978-3-642-34274-5_17

Cited by 12 publications

(5 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to reduce the size of the feature vector, the method selects a specified number of points at equal distance along the contours of the projections. Ballin et al [3] estimate the 3D optical flow related to the tracked people from point cloud data, summarising it by means of a 3D grid-based descriptor. Wang et al [22] fuse the skeleton information and a local occupancy pattern (LOP) based on the 3D point cloud around each joint.…”

Section: Based On Rgb-d Datamentioning

confidence: 99%

Fusion of Skeletal and Silhouette-Based Features for Human Action Recognition with RGB-D Devices

Chaaraoui

Padilla-López

Flórez-Revuelta

2013

2013 IEEE International Conference on Computer Vision Workshops

View full text Add to dashboard Cite

Section: Based On Rgb-d Datamentioning

confidence: 99%

Fusion of Skeletal and Silhouette-Based Features for Human Action Recognition with RGB-D Devices

Chaaraoui

Padilla-López

Flórez-Revuelta

2013

2013 IEEE International Conference on Computer Vision Workshops

View full text Add to dashboard Cite

“…Optical flow is also used more recently as input to deep nets, that are explored in detail in chapter 3. 3D scene flow [62,171], sometimes also referred to as 3D Optical flow, which uses data from spatial as well as depth information has also been proposed and used in action recognition [8].…”

Section: Optical Flowmentioning

confidence: 99%

“…where for a given point m we considered the value corresponding to its maximum shearlet coefficient and its associated shearing parameter k max ; Θ k max is the associated direction obtained using (8) and n is…”

Section: Fig 8cmentioning

confidence: 99%

Cross-View Action Recognition

2021

Computer Vision

View full text Add to dashboard Cite

Cross View Action Recognition (CVAR) appraises a system's ability to recognise actions from viewpoints that are unfamiliar to the system. The state of the art methods that train on large amounts of training data rely on variation in the training data itself to increase their ability to tackle viewpoints changes. Therefore, these methods not only require a large scale dataset of appropriate classes for the application every time they train, but also correspondingly large amount of computation power for the training process leading to high costs, in terms of time, effort, funds and electrical energy. In this thesis, we propose a methodological pipeline that tackles change in viewpoint, training on small datasets and employing sustainable amounts of resources. Our method uses the optical flow input with a stream of a pre-trained model as-is to obtain a feature. Thereafter, this feature is used to train a custom designed classifier that promotes view-invariant properties. Our method only uses video information as input, in contrast to another set of methods that approach CVAR by using depth or pose input at the expense of increased sensor costs. We present a number of comparative analysis that aided the design of the pipelines, farther assessing the power of each component in the pipeline. The technique can also be adopted to existing, trained classifiers, with minimal fine-tuning, as this work demonstrates by comparing classifiers including shallow classifiers, deep pre-trained classifiers and our proposed classifier trained from scratch. Additionally, we present a set of qualitative results that promote our understanding of the relationship between viewpoints in the feature-space. i Francesca Odone has been a better PhD advisor and mentor to me than anyone can ever hope to have, and I would like to convey my deepest gratitude for her time, effort, guidance and patience and most importantly, for always keeping her door open for her students. I would also like to thank Nicoletta Noceti for her guidance and our collaborations during the PhD study. I would like to thank University of Genova, DIBRIS, and previous and current colleagues for all the help and plenty of constructive conversations throughout the PhD term with special mention to Damiano Malafronte.My family has been remarkably supportive and encouraging. I would like to thank my mother, Sangeeta Goyal, who calls me her strength but in reality she is mine, my sister, Yamini Goyal, to whom I owe, in big part, my sanity and my father, Subhash Goyal. I could not have done this without them.

show abstract

“…Following the work of Shotton et al [19], which introduced a robust real-time method for skeleton capturing with random forests, the usage of 3D skeleton data for activity recognition gained popularity in the past few years [14], [16]. Machine learning methods that were applied to activity recognition include both discriminative classifiers such as support vector machines, k-nearest neighbours, and random forests [1], [2], [4], [25] as well as generative approaches such as Hidden Markov Models (HMMs). In our work, we use HMMs for modelling and recognising activity sequences.…”

Section: Related Workmentioning

confidence: 99%

Human activity recognition in the context of industrial human-robot interaction

Roitberg

Perzylo

Somani

et al. 2014

Signal and Information Processing Association Annual Summit and Conference (APSIPA), 2014 Asia-Pacific

View full text Add to dashboard Cite

Human activity recognition is crucial for intuitive cooperation between humans and robots. We present an approach for activity recognition for applications in the context of humanrobot interaction in industrial settings. The approach is based on spatial and temporal features derived from skeletal data of human workers performing assembly tasks. These features were used to train a machine learning framework, which classifies discrete time frames with Random Forests and subsequently models temporal dependencies between the resulting states with a Hidden Markov Model. We considered the following three groups of activities: Movement, Gestures, and Object handling. A dataset has been collected which is comprised of 24 recordings of several human workers performing such activities in a human-robot interaction environment, as typically seen at small and medium-sized enterprises. The evaluation shows that the approach achieves a recognition accuracy of up to 88% for some activities and an average accuracy of 73%.

show abstract

Human Action Recognition from RGB-D Frames Based on Real-Time 3D Optical Flow Estimation

Cited by 12 publications

References 21 publications

Fusion of Skeletal and Silhouette-Based Features for Human Action Recognition with RGB-D Devices

Fusion of Skeletal and Silhouette-Based Features for Human Action Recognition with RGB-D Devices

Cross-View Action Recognition

Human activity recognition in the context of industrial human-robot interaction

Contact Info

Product

Resources

About