Efficient Human Activity Classification from Egocentric Videos Incorporating Actor-Critic Reinforcement Learning

Lu, Yantao; Li, Yilan; Velipasalar, Senem

doi:10.1109/icip.2019.8803823

Cited by 11 publications

(12 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep reinforcement learning is a group of RL techniques that use deep neural networks, which are powerful function approximators, to deal with high dimensional action/state spaces [Li, 2017;Arulkumaran et al, 2017;Franc ¸ois-Lavet et al, 2018]. Deep RL has proven to be very successful, as demonstrated by [Mnih et al, 2013;Mnih et al, 2015;Jaderberg et al, 2019;Nguyen et al, 2017]. There are two main categories of DRL techniques: methods based on value-function and methods based on policy gradient.…”

Section: Deep Reinforcement Learningmentioning

confidence: 99%

“…Deep Q-Network (DQN) is one of the most popular methods in this category, which involves incorporating deep learning into the traditional Q-learning. DQN has achieved superhuman results in some Atari games [Mnih et al, 2013]. On the other hand, policy gradient methods parameterize the policy and try to directly optimize it with respect to the expected reward.…”

Section: Deep Reinforcement Learningmentioning

confidence: 99%

“…As is mentioned in [Lu et al, 2019], there are high resolution video data in many cases which demand lots of computational resources to be processed. To address this problem in the domain of activity recognition by egocentric video, a DRL-based approach is presented in [Lu et al, 2019], which aims at finding the region of interest (ROI) in each frame of the video. The problem of finding ROI can also be considered as a spatial attention finding problem.…”

Section: Spatial Attention Findingmentioning

confidence: 99%

“…DRL enables RL agents to make decisions using the unstructured input data without manually engineering the state space. It is mainly used to deal with high dimensional state and action spaces, which are prevalent in most of the practical real world problems such as video game playing [Torrado et al, 2018] and self-driving cars control [Sallab et al, 2017].…”

Section: Introductionmentioning

confidence: 99%

“…The progress and success of deep reinforcement learning methods has encouraged researchers to use it in computer vision applications in order to address the problems of large state and action spaces, which cannot be addressed with traditional RL methods. There are many DRL-based studies in the field of computer vision such as video captioning ], visual tracking [Yun et al, 2017, activity detection [Yeung et al, 2016], activity localization , face recognition [Rao et al, 2017], and person identification [Haque et al, 2016]. However, there are a few studies on DRL-based activity recognition.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Deep Reinforcement Learning in Human Activity Recognition: A Survey

Nikpour¹

2022

Preprint

View full text Add to dashboard Cite

<div>Human activity recognition is a popular research field in computer vision that has already been widely studied. However, it is still an active research field since it plays an important role in many current and emerging real world intelligent systems, like visual surveillance and human-computer interaction. Deep Reinforcement Learning (DRL) has recently been employed to address the activity recognition problem with various purposes, such as finding attention in video data or obtaining the best network structure. DRL-based human activity recognition has only been around for a short time, and it is a challenging, novel field[ of study. Therefore, to facilitate further research in this field, we have constructed a comprehensive survey on activity recognition methods that incorporate deep reinforcement learning. Towards the end of this survey, we summarize key challenges and open problems in this area that can be addressed by researchers in the future.</div>

show abstract

Section: Deep Reinforcement Learningmentioning

confidence: 99%

Section: Deep Reinforcement Learningmentioning

confidence: 99%

Section: Spatial Attention Findingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Deep Reinforcement Learning in Human Activity Recognition: A Survey

Nikpour¹

2022

Preprint

View full text Add to dashboard Cite

show abstract

Retrieval of Surgical Phase Transitions Using Reinforcement Learning

Zhang

Bano

Page

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

In minimally invasive surgery, surgical workflow segmentation from video analysis is a well studied topic. The conventional approach defines it as a multi-class classification problem, where individual video frames are attributed a surgical phase label. We introduce a novel reinforcement learning formulation for offline phase transition retrieval. Instead of attempting to classify every video frame, we identify the timestamp of each phase transition. By construction, our model does not produce spurious and noisy phase transitions, but contiguous phase blocks. We investigate two different configurations of this model. The first does not require processing all frames in a video (only < 60% and < 20% of frames in 2 different applications), while producing results slightly under the state-of-the-art accuracy. The second configuration processes all video frames, and outperforms the state-of-the art at a comparable computational cost. We compare our method against the recent topperforming frame-based approaches TeCNO and Trans-SVNet on the public dataset Cholec80 and also on an in-house dataset of laparoscopic sacrocolpopexy. We perform both a frame-based (accuracy, precision, recall and F1-score) and an event-based (event ratio) evaluation of our algorithms.

show abstract