When incorporating deep neural networks into robotic systems, a major challenge is the lack of uncertainty measures associated with their output predictions. Methods for uncertainty estimation in the output of deep object detectors (DNNs) have been proposed in recent works, but have had limited success due to 1) information loss at the detectors nonmaximum suppression (NMS) stage, and 2) failure to take into account the multitask, many-to-one nature of anchor-based object detection. To that end, we introduce BayesOD, an uncertainty estimation approach that reformulates the standard object detector inference and Non-Maximum suppression components from a Bayesian perspective. Experiments performed on four common object detection datasets show that BayesOD provides uncertainty estimates that are better correlated with the accuracy of detections, manifesting as a significant reduction of 9.77%-13.13% on the minimum Gaussian uncertainty error metric and a reduction of 1.63%-5.23% on the minimum Categorical uncertainty error metric. Code will be released at https://github.com/asharakeh/bayes-od-rc.

BayesOD: A Bayesian Approach for Uncertainty Estimation in Deep Object Detectors

When incorporating deep neural networks into robotic systems, a major challenge is the lack of uncertainty measures associated with their output predictions. Methods for uncertainty estimation in the output of deep object detectors (DNNs) have been proposed in recent works, but have had limited success due to 1) information loss at the detectors non-maximum suppression (NMS) stage, and 2) failure to take into account the multitask, many-to-one nature of anchor-based object detection. To that end, we introduce BayesOD, an uncertainty estimation approach that reformulates the standard object detector inference and Non-Maximum suppression components from a Bayesian perspective. Experiments performed on four common object detection datasets show that BayesOD provides uncertainty estimates that are better correlated with the accuracy of detections, manifesting as a significant reduction of 9.77\%-13.13\% on the minimum Gaussian uncertainty error metric and a reduction of 1.63\%-5.23\% on the minimum Categorical uncertainty error metric. Code will be released at {\url{this https URL}}.

BayesOD: A Bayesian Approach for Uncertainty Estimation in Deep Object Detectors.

Conventional surveillance systems such as CCTV still have limitations that merely viewing and recording. This limitation causes its function to only be passive monitoring and unable to provide real-time early warning systems as an effort to anticipate security threats or violations of regulations. The increasing need in the security sector, especially in public area, requires a solution in the form of a system that can detect suspicious objects through video surveillance systems. The integration of artificial intelligent, machine learning, image processing and computer vision become the latest study in surveillance system development innovation. Although there are many datasets, methods and frameworks available in previous research, there are still few papers that discuss the use of intelligent video analytics in detecting suspicious objects. This paper will comprehensively and systematically review the literature on applying machine learning for object detection and video surveillance systems published between 2010 and 2020. The literature extraction process is carried out by identifying and analyzing papers to describe the scope of research to detect suspicious objects using intelligent video analytics, frameworks, methods, datasets and identifying suspicious characteristics. At the end of this paper, conclusions have been outlined regarding the challenges and opportunities for suspicious object detection research using video analytics in the future.

Intelligent Video Analytic for Suspicious Object Detection : A Systematic Review

Inventory management in an automated industrial environment is the foremost requirement in order to shorten the gap between demand and supply. It comprises of object identification, localization, and counting. This paper introduces an approach for object counting in automated inventory management using KAZE features under different lighting conditions. Firstly, the prototype image and real-time inventory feed as the scene image are captured for detection of KAZE features. The detected features in the prototype image are subjected to density based scanning clustering algorithm. The KAZE features of each cluster obtained in the prototype image are mapped with the KAZE features of inventory feed. The mapped features in inventory feed are again subjected to density based scanning clustering algorithm. The clusters obtained in the inventory feed are then processed by Homography transform. Homography transform generates the predictions for object locations by projecting prototype corners in the inventory feed. The Homography transform projection results in rectangular box polygons in the inventory feed for the tentative location of prototype instances. Since there may be multiple predictions for a single object instance, the predicted object locations are integrated by density based scanning clustering algorithm to the centroids of these rectangular box polygons. It provides the exact location of prototype instances. Finally, the count is obtained. The graphical user interface for inventory management is also designed which exhibits user-friendly attributes. The proposed approach has also been compared with the previously developed approaches of object counting in automated inventory management. The experimental results state that the proposed approach outperforms the existing ones in the presence of different lighting conditions such as low-light or dim-light and bright light.

Object Counting using KAZE Features Under Different Lighting Conditions for Inventory Management

Spatial data classification and extraction is a significant problem to be resolved in data mining. The classification performance of existing techniques was not effectual to accurately mine the interesting spatial data. Furthermore, the amount of time taken for classifying the spatial data location was very higher. In order to resolve the above limitations, an Exponentiated Pareto based linear assorted classification method is introduced to reduce the incorrect classification of spatial data. This paper begins with a discussion of traditional methods of spatial data mining. The algorithm makes an association rule with spatial data objects. The technique conducts the experimental works using metrics such as classification accuracy, time complexity, space complexity and false positive rate with respect to different number of data. The proposed approach takes the forest fire dataset and El Nino dataset as input and predicts the burned area of forest fires and weather and climate conditions. The experimental results show that the proposed technique is able to increase the classification accuracy and reduce the time complexity as well as space complexity and false positive rate of spatial data mining as compared to state-of-the-art works.

A novel linear assorted classification method based association rule mining with spatial data

Now a day security is a major issue in public places such as railway stations, airports, shopping malls and in crowded areas. Unattended object detection is one of essential tasks in video surveillance system. This paper proposes an algorithm for detecting unattended and unknown object using background subtraction and morphological filtering. The main goal of the algorithm is to automatically detect the suspicious object and to improve the safety and security of the public areas. The system takes frames from video which is captured by a static camera as input and subtraction of foreground and background is done using thresholding techniques. To improve the detected region, morphological operation is used in the process. The system recognizes the reasons of interest, creates blobs of objects, labels the blobs and finally gives a warning when suspicious objects are detected.

Suspicious object detection in surveillance videos for security applications

http://gvschoolpub.org/journals/IJSTSD/vol2_no1/3.pdf

Suspicious Object Detection in Surveillance Videos for Security Applications

This paper presents a rate control algorithm for H.264 Scalable video codec, based on an independent rate control (IRC) scheme for spatial scalable layers. An independent bit allocation scheme was targeted for the impact of spatial scalable layer on bit budget achievements. For each spatial layer, the target bits for the current basic unit is a function of budget allocation for current Group of Pictures (GOP), predefined frame rate, target buffer level, buffer occupancy, and the available channel bandwidth . The allocated bits for a basic unit are updated such that the target bits for each basic unit adapt to the respective spatial scalable layer. This dynamic bit allocation scheme is implemented according to the complexity of a basic unit in each spatial layer. The proposed algorithm introduces more precise bit allocation within spatial scalable layers with defined buffer size. In a single framework, IRC offers control for GOP level, frame level and macro block (MB) level. At MB level IRC provides finer bit rate and buffer control. Experimental results demonstrate that compared to algorithm in Scalable video codec reference software , the proposed method provides faster settling of target bit rate and thus is very desirable for H.264/SVC rate control applications.

Independent rate control scheme for spatial scalable layer in H.264SVC

Scalable video coding extension has been added to H.264AVC to support compression and encoding of multiple resolution video sequences, having different frame rates and fidelities in a single bit stream. The motion vectors and the residual data of the enhancement layers are derived from up-sampling the co-located macroblock (MB) of the base layer. The peak signal to noise ratio (PSNR) across the enhancement layers is degraded as up-sampling introduces distortion of high-frequency components. In this paper, a spatial-resolution-ratio-based MB mode decision scheme is proposed for spatially enhanced layers. The scheme uses the motion estimated at the base layer, to encode the respective MBs in the enhancement layers. The spatial–temporal search schemes at the enhancement layers are used to derive motion vectors and residues that are encoded using a quantization parameter obtained using independent rate control (IRC) scheme. The IRC from the prior art is modified to achieve better rate control per layer by recursive updates for mean absolute difference values of each basic unit. Proposed modified inter-layer dependency shows improvement in the PSNR for enhancement layers while the updated IRC enforces better IRC for all the layers.

/pdf/efficient-rate-control-scheme-using-modified-inter-layer-1htaopfu31.pdf

Efficient rate control scheme using modified inter-layer dependency for spatial scalability

Every robotic arm has a specific work envelope and a predefined trajectory that it must follow to complete any task. Geometric calculation for determining joint angles and velocity involves complex trigonometric calculations and derivations which are not simple enough to actuate in a control software. Many times, geometric calculation also tend to become unsolvable due to complex work envelopes. Though iterative algorithms provide generalized solution to every serial link manipulator possible, it is computationally very expensive. If the number of nodes is large, this further aggravates the situation. The paper focuses on the methodology to combine iterative and geometrical methods to find inverse kinematics solutions to a 4 Degree of Freedom (DOF) robotic arm. Thus, the paper aims at establishing a bargain between geometric method complexities and iterative method computational load. Iterative method is employed for certain links and geometry is applied for rest of the links to achieve the desired end location with minimum error possible while adhering to the constraints of the work envelope.

P. M. Jadhav

Papers

Suspicious object detection in surveillance videos for security applications

Suspicious Object Detection in Surveillance Videos for Security Applications

Independent rate control scheme for spatial scalable layer in H.264SVC

Efficient rate control scheme using modified inter-layer dependency for spatial scalability

Inverse Kinematics Solution for a Robotic Arm Through Geometric and Iterative Fusion Based Modelling