Actor–Critic Reinforcement Learning and Application in Developing Computer-Vision-Based Interface Tracking

Dogru, Oguzhan; Kirubakaran, V.; Huang, Biao

doi:10.1016/j.eng.2021.04.027

Cited by 20 publications

(5 citation statements)

References 82 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Reinforcement learning is widely been used in many computer vision problems [21,30], where it has not been explored for the problem of VD in real-world environments. Reinforcement learning techniques have surely accomplished complex tasks with higher effective and efficient results.…”

Section: Reinforcement Learning For Vdmentioning

confidence: 99%

An overview of violence detection techniques: current challenges and future directions

et al. 2022

View full text Add to dashboard Cite

The Big Video Data generated in today's smart cities has raised concerns from its purposeful usage perspective, where surveillance cameras, among many others are the most prominent resources to contribute to the huge volumes of data, making its automated analysis a difficult task in terms of computation and preciseness. Violence Detection (VD), broadly plunging under Action and Activity recognition domain, is used to analyze Big Video data for anomalous actions incurred due to humans. The VD literature is traditionally based on manually engineered features, though advancements to deep learning based standalone models are developed for

show abstract

Section: Reinforcement Learning For Vdmentioning

confidence: 99%

An overview of violence detection techniques: current challenges and future directions

et al. 2022

View full text Add to dashboard Cite

show abstract

“…The primary separation vessel is a key component in the oil sands industry, where an interface between two liquids is the target variable to be detected and tracked accurately [53]. The tank is schematically shown in Fig.…”

Section: B Application 1: Online Reward Estimation For the Rl Agent I...mentioning

confidence: 99%

“…Because the RL agent tries to maximize the reward, r = x ≤ 0 and y ≤ 0 are often preferred to provide feasibility in control applications [35], [52], [53]. Due to this property, when the noisy reward magnitude is low, the resulting reward becomes a truncated variable [35].…”

Section: Objective Fcnmentioning

confidence: 99%

Skew Filtering for Online State Estimation and Control

DOGRU¹,

Chiplunkar²,

Huang³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

<div>Process optimization and control can become challenging when the measurements are affected by irregular noise. Classical approaches utilize Gaussian method like Kalman filtering to alleviate the sensory noise. However, many industries involve skewed noise in their processes. While the closed skew normal (CSN) distribution generalizes a Gaussian distribution with additional parameters, its dimension increases during recursive estimation, making it impractical. Even though there exist some techniques for the solution, they are typically either too complicated or inaccurate for the higher dimensional problems. This study proposes a novel online optimization scheme to reduce the dimensionality of a CSN distribution while considering the properties of the complete empirical distribution. Since the objective function used during the optimization step considers the geometry of the metric space, the proposed scheme achieves the higher accuracy without sacrificing computational efficiency. After finding the reliable combination of objective function and optimizer, the proposed filter is applied to two real-time pilot-scale experiments. The results indicate that it is beneficial for recursive state estimation in the presence of skewed noise.<br></div>

show abstract

“…Deep Reinforcement Learning (RL) based approaches 9 utilize a series of recurrent neural network layers to track objects in image data. Current actor critic based RL approaches to track objects in image data in real time utilizes an actor network with a continuous state space that is trained both online and offline 10 as well as a discrete action space 11 . Both methods performs well when compared with other state of the art approaches.…”

Section: Introductionmentioning

confidence: 99%

Adaptive critic network for person tracking using 3D skeleton data

Zalameda

Glandon

Iftekharuddin

2023

Pattern Recognition and Tracking XXXIV

View full text Add to dashboard Cite

Analysis of human gait using 3-dimensional co-occurrence skeleton joints extracted from Lidar sensor data has been shown a viable method for predicting person identity. The co-occurrence based networks rely on the spatial changes between frames of each joint in the skeleton data sequence. Normally, this data is obtained using a Lidar skeleton extraction method to estimate these co-occurrence features from raw Lidar frames, which can be prone to incorrect joint estimations when part of the body is occluded. These datasets can also be time consuming and expensive to collect and typically offer a small number of samples for training and testing network models. The small number of samples and occlusion can cause challenges when training deep neural networks to perform real time tracking of the person in the scene. We propose preliminary results with a deep reinforcement learning actor critic network for person tracking of 3D skeleton data using a small dataset. The proposed approach can achieve an average tracking rate of 68.92±15.90% given limited examples to train the network.

show abstract

Actor–Critic Reinforcement Learning and Application in Developing Computer-Vision-Based Interface Tracking

Cited by 20 publications

References 82 publications

An overview of violence detection techniques: current challenges and future directions

An overview of violence detection techniques: current challenges and future directions

Skew Filtering for Online State Estimation and Control

Adaptive critic network for person tracking using 3D skeleton data

Contact Info

Product

Resources

About