Scribd
Upload
372 views8 pages

Human Activity Recognization

Human activity recognition aims to recognize activities from observations of subjects' actions and environmental conditions. It has applications in video surveillance, healthcare, and human-computer interaction. Advances in this field could greatly benefit surveillance with video cameras. Deep learning techniques can extract high-level representations and are increasingly used for activity recognition, though lack of large datasets remains a challenge.

Uploaded by

NIck
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
372 views8 pages

Human Activity Recognization

Human activity recognition aims to recognize activities from observations of subjects' actions and environmental conditions. It has applications in video surveillance, healthcare, and human-computer interaction. Advances in this field could greatly benefit surveillance with video cameras. Deep learning techniques can extract high-level representations and are increasingly used for activity recognition, though lack of large datasets remains a challenge.

Uploaded by

NIck
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 8

Human Activity

Recognition
Neel Ponkia (17BCE164)
Milan Pethani ((17BCE163)
Outline

 Human activity recognition (HAR) aims to recognize activities from a series of


observations on the actions of subjects and the environmental conditions.
 The vision-based HAR research is the basis of many applications including
video surveillance, health care, and humancomputer interaction (HCI)
 Surveillance by video cameras could benefit greatly by advances in this field.
In the area of robotics, the tasks of autonomous navigation or social
interaction could also take advantage of the knowledge extracted from live
video recording.
Introduction

 Action recognition instinctively implies recognizing a set of actions. From a


standpoint of computer vision, action recognition alludes to learning a lot of
video arrangements to recognize the grouping of developments related with a
specific activity and utilizing the information acquired to foresee a future
activity dependent on the developments related.
 Action recognition is a key segment in numerous applications, to be specific,
human-computer interaction, surveillance, video analysis and so forth.
Recognizing an action encourages the framework to outline a whole occasion.
Introduction

 Hence, action recognition involves the classification of different actions from


videos, a sequence of frames, taking into account as well the fact that the
action could not be performed during the entire video.
 the progress for video classification has been slower due to various reasons:
 Apart from spatial information, temporal context across frames is also
required.
 Huge computational cost.
 Datasets are more limited, due to the difficulty to collect, annotate and store
videos.
Techniques and other tasks related to
this subject
Advantages

 There is no need of expert information to get appropriate features, lessening


the exertion of feature extraction.
 Rather than planning them physically, features are naturally learned through
the network.
 Deep neural networks can extract elevated level portrayal in deep layer,
making it progressively appropriate for complex undertakings.
Disadvantages

 Due to lack of datasets, we need to collect more and more data.


 It is very tedious task.
 Models have difficulty in generalization.
Conclusion

 Video-based activity recognition is more muddled than static image


classification and this is likewise reflected in the outcomes got up until now.
Be that as it may, since deep learning is as yet being exploited, soon, this
assignment may get simpler to perform and current outcomes might be
improved utilizing some deep learning procedures.

You might also like