-
Collecting Consistently High Quality Object Tracks with Minimal Human Involvement by Using Self-Supervised Learning to Detect Tracker Errors
Authors:
Samreen Anjum,
Suyog Jain,
Danna Gurari
Abstract:
We propose a hybrid framework for consistently producing high-quality object tracks by combining an automated object tracker with little human input. The key idea is to tailor a module for each dataset to intelligently decide when an object tracker is failing and so humans should be brought in to re-localize an object for continued tracking. Our approach leverages self-supervised learning on unlab…
▽ More
We propose a hybrid framework for consistently producing high-quality object tracks by combining an automated object tracker with little human input. The key idea is to tailor a module for each dataset to intelligently decide when an object tracker is failing and so humans should be brought in to re-localize an object for continued tracking. Our approach leverages self-supervised learning on unlabeled videos to learn a tailored representation for a target object that is then used to actively monitor its tracked region and decide when the tracker fails. Since labeled data is not needed, our approach can be applied to novel object categories. Experiments on three datasets demonstrate our method outperforms existing approaches, especially for small, fast moving, or occluded objects.
△ Less
Submitted 6 May, 2024;
originally announced May 2024.
-
VQA Therapy: Exploring Answer Differences by Visually Grounding Answers
Authors:
Chongyan Chen,
Samreen Anjum,
Danna Gurari
Abstract:
Visual question answering is a task of predicting the answer to a question about an image. Given that different people can provide different answers to a visual question, we aim to better understand why with answer groundings. We introduce the first dataset that visually grounds each unique answer to each visual question, which we call VQAAnswerTherapy. We then propose two novel problems of predic…
▽ More
Visual question answering is a task of predicting the answer to a question about an image. Given that different people can provide different answers to a visual question, we aim to better understand why with answer groundings. We introduce the first dataset that visually grounds each unique answer to each visual question, which we call VQAAnswerTherapy. We then propose two novel problems of predicting whether a visual question has a single answer grounding and localizing all answer groundings. We benchmark modern algorithms for these novel problems to show where they succeed and struggle. The dataset and evaluation server can be found publicly at https://vizwiz.org/tasks-and-datasets/vqa-answer-therapy/.
△ Less
Submitted 24 August, 2023; v1 submitted 21 August, 2023;
originally announced August 2023.
-
X-ray emission of radio-loud quasar SDSS J121426.52+140258.9: independent variations between optical/UV and X-ray emission
Authors:
Minhua Zhou,
Minfeng Gu,
Mai Liao,
Muhammad S. Anjum
Abstract:
To understand the X-ray emission of active galactic nuclei (AGNs), we explored the optical-to-X-ray variation correlation of a radio-loud quasar (RLQ) SDSS J121426.52+140258.9 (hereafter J1214+1402) with multi-epoch observations of Swift and XMM-Newton telescopes. With the historical multi-band data, we found that the infrared to X-ray flux of RLQ J1214+1402 should not be dominated by the beamed j…
▽ More
To understand the X-ray emission of active galactic nuclei (AGNs), we explored the optical-to-X-ray variation correlation of a radio-loud quasar (RLQ) SDSS J121426.52+140258.9 (hereafter J1214+1402) with multi-epoch observations of Swift and XMM-Newton telescopes. With the historical multi-band data, we found that the infrared to X-ray flux of RLQ J1214+1402 should not be dominated by the beamed jet emission. The Swift optical/UV and X-ray light curves showed that J1214+1402 has two optical states with low flux before 2014 April 08 and high flux after 2014 June 11, but has no significant X-ray variations during the time range between 2007 March 09 and 2014 August 04. This result was supported by the XMM-Newton observations in the overlapped time with Swift. Interestingly, the early XMM-Newton data prior to the Swift time presents two unusual emission epochs when J1214+1402 has relatively low optical fluxes but has the brightest X-ray fluxes. The overall independence of optical-to-X-ray variation seems hard to be described by the disk-corona model. With the X-ray spectral fitting, we find that the soft X-ray excess in J1214+1402 appears only during the high optical state when the X-ray emission is at low state. The soft X-ray excess in J1214+1402 is difficult to be explained by the ionized accretion disk, instead, it may be related to the warm corona.
△ Less
Submitted 23 November, 2022;
originally announced November 2022.
-
Blockchain associated machine learning and IoT based hypoglycemia detection system with auto-injection feature
Authors:
Rahnuma Mahzabin,
Fahim Hossain Sifat,
Sadia Anjum,
Al-Akhir Nayan,
Muhammad Golam Kibria
Abstract:
Hypoglycemia is an unpleasant phenomenon caused by low blood glucose. The disease can lead a person to death or a high level of body damage. To avoid significant damage, patients need sugar. The research aims at implementing an automatic system to detect hypoglycemia and perform automatic sugar injections to save a life. Receiving the benefits of the internet of things (IoT), the sensor data was t…
▽ More
Hypoglycemia is an unpleasant phenomenon caused by low blood glucose. The disease can lead a person to death or a high level of body damage. To avoid significant damage, patients need sugar. The research aims at implementing an automatic system to detect hypoglycemia and perform automatic sugar injections to save a life. Receiving the benefits of the internet of things (IoT), the sensor data was transferred using the hypertext transfer protocol (HTTP) protocol. To ensure the safety of health-related data, blockchain technology was utilized. The glucose sensor and smartwatch data were processed via Fog and sent to the cloud. A Random Forest algorithm was proposed and utilized to decide hypoglycemic events. When the hypoglycemic event was detected, the system sent a notification to the mobile application and auto-injection device to push the condensed sugar into the victims body. XGBoost, k-nearest neighbors (KNN), support vector machine (SVM), and decision tree were implemented to compare the proposed models performance. The random forest performed 0.942 testing accuracy, better than other models in detecting hypoglycemic events. The systems performance was measured in several conditions, and satisfactory results were achieved. The system can benefit hypoglycemia patients to survive this disease.
△ Less
Submitted 26 July, 2022;
originally announced August 2022.
-
The Curvature of Spectral Energy Distribution and $γ$-ray Dominance of Fermi BL Lac Objects
Authors:
Muhammad Shahzad Anjum,
Liang Chen,
Minfeng Gu
Abstract:
The extragalactic $γ$-ray sky is dominated by blazars and their study plays an important role in understanding jet physics, cosmic evolution history and origin of ultra high energy cosmic rays. In this work, we study a large sample of BL Lac objects to investigate why some sources are detected in $γ$-rays, while others not. We selected 170 BL Lac objects, with measured synchrotron spectral curvatu…
▽ More
The extragalactic $γ$-ray sky is dominated by blazars and their study plays an important role in understanding jet physics, cosmic evolution history and origin of ultra high energy cosmic rays. In this work, we study a large sample of BL Lac objects to investigate why some sources are detected in $γ$-rays, while others not. We selected 170 BL Lac objects, with measured synchrotron spectral curvature and Doppler factor, and divided them into Fermi-LAT detected (FBLs) and non-detected (NFBLs) sources. We show that FBLs have smaller curvature than NFBLs, even after getting rid of Doppler beaming effect. The BL Lac objects PKS 0048-09 and S5 0716+714 have similar synchrotron peak frequency and luminosity but different $γ$-ray dominance and their quasi-simultaneous broadband spectral energy distributions (SEDs) can be well fitted by a log-parabolic synchrotron self-Compton (SSC) model with same jet parameters except for the curvature and source size, assuming curvature being proportional to the size of emission region. Our results imply that for a given synchrotron luminosity, the different SED curvature and Compton dominance may account for the discrepancy between FBLs and NFBLs. We discuss these results in context of stochastic particle acceleration and radiation mechanisms.
△ Less
Submitted 20 May, 2022;
originally announced May 2022.
-
Grounding Answers for Visual Questions Asked by Visually Impaired People
Authors:
Chongyan Chen,
Samreen Anjum,
Danna Gurari
Abstract:
Visual question answering is the task of answering questions about images. We introduce the VizWiz-VQA-Grounding dataset, the first dataset that visually grounds answers to visual questions asked by people with visual impairments. We analyze our dataset and compare it with five VQA-Grounding datasets to demonstrate what makes it similar and different. We then evaluate the SOTA VQA and VQA-Groundin…
▽ More
Visual question answering is the task of answering questions about images. We introduce the VizWiz-VQA-Grounding dataset, the first dataset that visually grounds answers to visual questions asked by people with visual impairments. We analyze our dataset and compare it with five VQA-Grounding datasets to demonstrate what makes it similar and different. We then evaluate the SOTA VQA and VQA-Grounding models and demonstrate that current SOTA algorithms often fail to identify the correct visual evidence where the answer is located. These models regularly struggle when the visual evidence occupies a small fraction of the image, for images that are higher quality, as well as for visual questions that require skills in text recognition. The dataset, evaluation server, and leaderboard all can be found at the following link: https://vizwiz.org/tasks-and-datasets/answer-grounding-for-vqa/.
△ Less
Submitted 8 April, 2022; v1 submitted 4 February, 2022;
originally announced February 2022.
-
CrowdMOT: Crowdsourcing Strategies for Tracking Multiple Objects in Videos
Authors:
Samreen Anjum,
Chi Lin,
Danna Gurari
Abstract:
Crowdsourcing is a valuable approach for tracking objects in videos in a more scalable manner than possible with domain experts. However, existing frameworks do not produce high quality results with non-expert crowdworkers, especially for scenarios where objects split. To address this shortcoming, we introduce a crowdsourcing platform called CrowdMOT, and investigate two micro-task design decision…
▽ More
Crowdsourcing is a valuable approach for tracking objects in videos in a more scalable manner than possible with domain experts. However, existing frameworks do not produce high quality results with non-expert crowdworkers, especially for scenarios where objects split. To address this shortcoming, we introduce a crowdsourcing platform called CrowdMOT, and investigate two micro-task design decisions: (1) whether to decompose the task so that each worker is in charge of annotating all objects in a sub-segment of the video versus annotating a single object across the entire video, and (2) whether to show annotations from previous workers to the next individuals working on the task. We conduct experiments on a diversity of videos which show both familiar objects (aka - people) and unfamiliar objects (aka - cells). Our results highlight strategies for efficiently collecting higher quality annotations than observed when using strategies employed by today's state-of-art crowdsourcing system.
△ Less
Submitted 29 September, 2020;
originally announced September 2020.
-
On the Origin and Evolution of Curvature of the Spectral Energy Distribution of Fermi Bright Blazars
Authors:
Muhammad S. Anjum,
Liang Chen,
Minfeng Gu
Abstract:
The origin and evolution of spectral curvature in blazar spectral energy distribution (SED) is still unclear. Since the observed SED curvature is related to an intrinsic curvature in emitting electron energy distribution (EED), we study this question by employing a log-parabolic EED with a curvature parameter and peak energy to model the quasi-simultaneous broadband SEDs of selected blazars in Fer…
▽ More
The origin and evolution of spectral curvature in blazar spectral energy distribution (SED) is still unclear. Since the observed SED curvature is related to an intrinsic curvature in emitting electron energy distribution (EED), we study this question by employing a log-parabolic EED with a curvature parameter and peak energy to model the quasi-simultaneous broadband SEDs of selected blazars in Fermi-LAT Bright AGN Sample (LBAS) using synchrotron and inverse Compton (IC) processes. We find that log-parabolic IC model can successfully explain the emission in all blazars in our sample. On average, FSRQs have higher magnetic field, Doppler factor, and curvature than BL Lac objects. The BL Lac objects show an anticorrelation between the curvature parameter of the EED and its peak energy, which is a signature of stochastic acceleration. FSRQs do not manifest such correlation and rather show a mild positive relationship between these parameters. This suggests that the evolution of spectral curvature in the BL Lac objects is dominated by a strong stochastic acceleration component, whereas the curvature in FSRQs evolves in a cooling dominated regime due to an additional external Compton (EC) component. The strong cooling in FSRQs not only restricts the electron peak energy but also adds extra curvature to the high energy tail of emitting EED. Since the curvature decreases from FSRQs toward high peak BL Lac objects (HBLs), opposite to peak energy, the curvature parameter can be considered a third parameter of the blazar sequence in addition to peak frequency and luminosity.
△ Less
Submitted 1 July, 2020; v1 submitted 4 June, 2020;
originally announced June 2020.
-
Symmetric mixtures of pusher and puller microswimmers behave as noninteracting suspensions
Authors:
Dóra Bárdfalvy,
Shan Anjum,
Cesare Nardini,
Alexander Morozov,
Joakim Stenhammar
Abstract:
Suspensions of rear- and front-actuated microswimmers immersed in a fluid, known respectively as ``pushers'' and ``pullers'', display qualitatively different collective behaviours: beyond a characteristic density, pusher suspensions exhibit a hydrodynamic instability leading to collective motion known as active turbulence, a phenomenon which is absent for pullers. In this Letter, we describe the c…
▽ More
Suspensions of rear- and front-actuated microswimmers immersed in a fluid, known respectively as ``pushers'' and ``pullers'', display qualitatively different collective behaviours: beyond a characteristic density, pusher suspensions exhibit a hydrodynamic instability leading to collective motion known as active turbulence, a phenomenon which is absent for pullers. In this Letter, we describe the collective dynamics of a binary pusher--puller mixture using kinetic theory and large-scale particle-resolved simulations. We derive and verify an instability criterion, showing that the critical density for active turbulence moves to higher values as the fraction $χ$ of pullers is increased and disappears for $χ\geq 0.5$. We then show analytically and numerically that the two-point hydrodynamic correlations of the 1:1 mixture are equal to those of a suspension of noninteracting swimmers. Strikingly, our numerical analysis furthermore shows that the full probability distribution of the fluid velocity fluctuations collapses onto the one of a noninteracting system at the same density, where swimmer--swimmer correlations are strictly absent. Our results thus indicate that the fluid velocity fluctuations in 1:1 pusher--puller mixtures are exactly equal to those of the corresponding noninteracting suspension at any density, a surprising cancellation with no counterpart in equilibrium long-range interacting systems.
△ Less
Submitted 7 July, 2020; v1 submitted 15 April, 2020;
originally announced April 2020.
-
Electromagnetic source localization with finite set of frequency measurements
Authors:
Abdul Wahab,
Amer Rasheed,
Rab Nawaz,
Saman Anjum
Abstract:
A phase conjugation algorithm for localizing an extended radiating electromagnetic source from boundary measurements of the electric field is presented. Measurements are taken over a finite number of frequencies. The artifacts related to the finite frequency data are tackled with $l_1-$regularization blended with the fast iterative shrinkage-thresholding algorithm with backtracking of Beck & Tebou…
▽ More
A phase conjugation algorithm for localizing an extended radiating electromagnetic source from boundary measurements of the electric field is presented. Measurements are taken over a finite number of frequencies. The artifacts related to the finite frequency data are tackled with $l_1-$regularization blended with the fast iterative shrinkage-thresholding algorithm with backtracking of Beck & Teboulle.
△ Less
Submitted 16 September, 2014; v1 submitted 20 March, 2014;
originally announced March 2014.
-
Object Oriented Model for Evaluation of On-Chip Networks
Authors:
Sheraz Anjum,
Ehsan Ullah Munir,
Waqas Anwar,
Nadeem Javaid
Abstract:
The Network on Chip (NoC) paradigm is rapidly replacing bus based System on Chip (SoC) designs due to their inherent disadvantages such as non-scalability, saturation and congestion. Currently very few tools are available for the simulation and evaluation of on-chip architectures. This study proposes a generic object oriented model for performance evaluation of on-chip interconnect architectures a…
▽ More
The Network on Chip (NoC) paradigm is rapidly replacing bus based System on Chip (SoC) designs due to their inherent disadvantages such as non-scalability, saturation and congestion. Currently very few tools are available for the simulation and evaluation of on-chip architectures. This study proposes a generic object oriented model for performance evaluation of on-chip interconnect architectures and algorithms. The generic nature of the proposed model can help the researchers in evaluation of any kind of on-chip switching networks. The model was applied on 2D-Mesh and 2D-Diagonal-Mesh on-chip switching networks for verification and selection of best out of both the analyzed architectures. The results show the superiority of 2D-Diagonal-Mesh over 2D-Mesh in terms of average packet delay.
△ Less
Submitted 23 March, 2013;
originally announced April 2013.
