Hayder Radha - Profile on Academia.edu (original) (raw)
Papers by Hayder Radha
The area of domain adaptation has been instrumental in addressing the domain shift problem encoun... more The area of domain adaptation has been instrumental in addressing the domain shift problem encountered by many applications. This problem arises due to the difference between the distributions of source data used for training in comparison with target data used during realistic testing scenarios. In this paper, we introduce a novel MultiScale Domain Adaptive YOLO (MS-DAYOLO) framework that employs multiple domain adaptation paths and corresponding domain classifiers at different scales of the recently introduced YOLOv4 object detector to generate domain-invariant features. We train and test our proposed method using popular datasets. Our experiments show significant improvements in object detection performance when training YOLOv4 using the proposed MS-DAYOLO and when tested on target data representing challenging weather conditions for autonomous driving applications.
In this paper, we analyze statistical and rate-distortion (R-D) properties of MPEG-4 Fine-Granula... more In this paper, we analyze statistical and rate-distortion (R-D) properties of MPEG-4 Fine-Granular Scalability (FGS), which has recently become an important scalable compression framework and a de-facto standard for Internet video streaming. We first propose a novel statistical model of DCT residue that accurately captures the properties of the input to the MPEG-4 FGS enhancement layer. Our results show that FGS residue concentrates a lot of probability mass near zero and cannot be accurately modeled by Gaussian or Laplacian distributions. We then model the distortion of each bitplane based on the proposed statistical framework and further demonstrate that our R-D model significantly outperforms current distortion models.
The convergence of the Internet with new wireless and mobile networks is creating a whole new lev... more The convergence of the Internet with new wireless and mobile networks is creating a whole new level of heterogeneity in multimedia communications. This increased level of heterogeneity emphasizes the need for scalable and adaptive video solutions both for coding and transmission purposes. However, in general, there is an inherent tradeoff between the level of scalability and the quality of scalable video streams. In other words, the higher the bandwidth variation, the lower the overall video quality of the scalable stream that is needed to support the desired bandwidth range. In this paper, we introduce the notion of TranScaling (TS) which is a generalization of (non-scalable) transcoding. With transcaling, a scalable video stream, that covers a given bandwidth range, is mapped into one or more scalable video streams covering different bandwidth ranges. Our proposed TS framework exploits the fact that the level of heterogeneity changes at different points of the video distribution tree over wireless and mobile Internet networks. This provides the opportunity to improve the video quality by performing the appropriate transcaling process. We argue that an Internet/wireless network gateway represents a good candidate for performing transcaling. Moreover, we describe Hierarchical TranScaling (HTS) which provides a "Transcalar" the option of choosing among different levels of transcaling processes with different complexities. We illustrate the benefits of transcaling by considering the recently developed MPEG-4 Fine-Granular-Scalability (FGS) video coding. Simulation results of video transcaling are also presented.
Radar Tracking With Orthogonal Velocity Measurements for Autonomous Ground Vehicles
In this work we examine the effect of having independent radial and angular velocity radar measur... more In this work we examine the effect of having independent radial and angular velocity radar measurements on 2D tracking performance. With this, we provide design considerations for systems using interferometric angular velocity estimation. We found that on average as the number of clutter detections increases, the reduction in error by adding an additional angular velocity measurement also increases, implying a greater need for orthogonal velocity measurements in high-clutter environments. In particular, in simulation at a clutter rate of 50 detections per measurement, we obtained an error reduction of 14% by adding a radial velocity measurement and a further reduction of 3% from an additional angular velocity measurement. We also tested this method on a publicly available autonomous driving dataset by synthesizing angular velocity measurements and found a similar reduction of 13% and 6% by adding Doppler and angular velocity measurements, respectively.
Multimedia Over Wireless
CRC Press eBooks, Mar 22, 2000
Notice of Removal Hyperspectral material classification under monochromatic and trichromatic sampling rates
3D Multi-Object Tracking using Random Finite Set-based Multiple Measurement Models Filtering (RFS-M3) for Autonomous Vehicles
Multiple object tracking (MOT) is a critical module for enabling autonomous vehicles to achieve s... more Multiple object tracking (MOT) is a critical module for enabling autonomous vehicles to achieve safe planing and navigation in cluttered environments. In tracking-by-detection systems, there are inevitably many false positives and misses among learning-based input detections. The challenge for MOT is to combine these detections into tracks, and filter them based on their uncertainties, states, and temporal consistency to achieve accurate and persistent tracks. In this paper, we propose to solve the 3D MOT problem for autonomous driving applications using a random finite set-based (RFS) Multiple Measurement Models filter (RFS-M3). In partiuclar, we propose multiple measurement models for a Poisson multi-Bernoulli mixture (PMBM) filter in support of different application scenarios. Our RFS-M3 filter can naturally model these uncertainties accurately and elegantly. We combine the learning-based detections with our RFS-M3 tracker through incorporating the detection confidence score into the PMBM prediction and update step. The superior experimental results of our RFS-M3 tracker on Waymo, Argoverse and nuSceness datasets illustrate that our RFS-M3 tracker outperforms state-of-the-art deep learning-based and traditional filter-based approaches. To the best of our knowledge, this represents a first successful attempt for employing an RFS-based approach in conjunction with 3D learning-based amodal detections for 3D MOT applications with comprehensive validation using challenging datasets made available by industry leaders.
Under traditional IP multicast, application-level FEC can only be implemented on an end-to-end ba... more Under traditional IP multicast, application-level FEC can only be implemented on an end-to-end basis between the sender and the clients. Emerging overlay and peer-to-peer (p2p) networks open the door for new paradigms of network FEC. The deployment of FEC within these emerging networks has received very little attention (if any). In this paper, we analyze and optimize the impact of Network-Embedded FEC (NEF) in overlay and p2p multimedia multicast networks. Under NEF, we place FEC codecs in selected intermediate nodes of a multicast tree. The NEF codecs detect and recover lost packets within FEC blocks at earlier stages before these blocks arrive at deeper intermediate nodes or at the final leaf nodes. This approach significantly reduces the probability of receiving undecodable FEC blocks. In essence, the proposed NEF codecs work as signal regenerators in a communication system and can reconstruct most of the lost data packets without requiring retransmission. We develop an optimization algorithm for the placement of NEF codecs within random multicast trees. Our theoretical analysis and simulation results show that a relatively small number of NEF codecs placed in (sub-)optimally selected intermediate nodes of a network can improve the throughput and overall reliability dramatically.
arXiv (Cornell University), Sep 1, 2020
There have been significant advances in neural networks for both 3D object detection using LiDAR ... more There have been significant advances in neural networks for both 3D object detection using LiDAR and 2D object detection using video. However, it has been surprisingly difficult to train networks to effectively use both modalities in a way that demonstrates gain over single-modality networks. In this paper, we propose a novel Camera-LiDAR Object Candidates (CLOCs) fusion network. CLOCs fusion provides a low-complexity multi-modal fusion framework that significantly improves the performance of single-modality detectors. CLOCs operates on the combined output candidates before Non-Maximum Suppression (NMS) of any 2D and any 3D detector, and is trained to leverage their geometric and semantic consistencies to produce more accurate final 3D and 2D detection results. Our experimental evaluation on the challenging KITTI object detection benchmark, including 3D and bird's eye view metrics, shows significant improvements, especially at long distance, over the state-of-the-art fusion based methods. At time of submission, CLOCs ranks the highest among all the fusion-based methods in the official KITTI leaderboard. We will release our code upon acceptance.
Integrated Generative-Model Domain-Adaptation for Object Detection under Challenging Conditions
2022 IEEE 95th Vehicular Technology Conference: (VTC2022-Spring), Jun 1, 2022
Poster Abstract: Wind Speed and Direction Estimation Using Manifold Approximation
ABSTRACT
Analysis and design of reliable and stable link-layer protocols for wireless communication
ABSTRACT
Rain-Adaptive Intensity-Driven Object Detection for Autonomous Vehicles
SAE technical paper series, Apr 14, 2020
arXiv (Cornell University), May 7, 2014
arXiv (Cornell University), Jun 30, 2020
Advanced automotive active-safety systems, in general, and autonomous vehicles, in particular, re... more Advanced automotive active-safety systems, in general, and autonomous vehicles, in particular, rely heavily on visual data to classify and localize objects such as pedestrians, traffic signs and lights, and other nearby cars, to assist the corresponding vehicles maneuver safely in their environments. However, the performance of object detection methods could degrade rather significantly under challenging weather scenarios including rainy conditions. Despite major advancements in the development of deraining approaches, the impact of rain on object detection has largely been understudied, especially in the context of autonomous driving. The main objective of this paper is to present a tutorial on state-of-the-art and emerging techniques that represent leading candidates for mitigating the influence of rainy conditions on an autonomous vehicle's ability to detect objects. Our goal includes surveying and analyzing the performance of object detection methods trained and tested using visual data captured under clear and rainy conditions. Moreover, we survey and evaluate the efficacy and limitations of leading deraining approaches, deep-learning based domain adaptation, and image translation frameworks that are being considered for addressing the problem of object detection under rainy conditions. Experimental results of a variety of the surveyed techniques are presented as part of this tutorial.
PEEC: A Channel-Adaptive Feedback-Based Error Control Protocol for Wireless MAC Layer
Abstract—Reliable transmission is a challenging task over wireless LANs since wireless links are ... more Abstract—Reliable transmission is a challenging task over wireless LANs since wireless links are known to be susceptible to errors. Although the current IEEE802.11 standard ARQ error control protocol performs relatively well over channels with very low bit error rates (BERs), this performance deteriorates rapidly as the BER increases. This paper investigates the problem of reliable transmission in a contention free wireless LAN and introduces a Packet Embedded Error Control (PEEC) protocol, which employs packet-embedded parity symbols instead of ARQbased retransmission for error recovery. Specifically, depending on receiver feedback, PEEC adaptively estimates channel conditions and administers the transmission of (data and parity) symbols within a packet. This enables successful recovery of both new data and old unrecovered data from prior transmissions. In addition to theoretically analyzing PEEC, the performance of the proposed scheme is extensively analyzed over real channel traces collected on 802.11b WLANs. We compare PEEC performance with the performance of the IEEE802.11 standard ARQ protocol as well as contemporary protocols such as enhanced ARQ and the hybrid ARQ/FEC. Our analysis and experimental simulations show that PEEC outperforms all three competing protocols over a wide range of actual 802.11b WLAN collected traces. Finally, the design and implementation of PEEC using an Adaptive Low-Density-Parity-Check (A-LDPC) decoder is presented. Index Terms—Forward error correction, Channel coding, Wireless LAN, Feedback communication, Automatic repeat request.
Real-time switching of MPEG-2 bitstreams
ABSTRACT
Fast image super-resolution via selective manifold learning of high-resolution patches
This paper considers the problem of single image super-resolution (SR). Previous example-based SR... more This paper considers the problem of single image super-resolution (SR). Previous example-based SR approaches mainly focus on analyzing the co-occurrence properties of low resolution (LR) and high resolution (HR) patches via dictionary learning. In our recent work [1], a novel approach (SR via sparse subspace clustering-based linear approximation of manifold or SLAM) has been proposed. In this paper, we further improve the SLAM method by considering and analyzing each tangent subspace as one point in a Grassmann manifold to select an optimal subset of tangent spaces. Furthermore, the optimal subset is clustered hierarchically, which helps in reducing the proposed algorithm's complexity significantly while still preserving the quality of the reconstructed HR image.
The contourlet transform is a new directional transform, which is capable of capturing contours a... more The contourlet transform is a new directional transform, which is capable of capturing contours and fine details in images. We recently introduced the wavelet-based contourlet transform (WBCT) that is a non-redundant version of the contourlet transform, and appropriately used this transform for image coding. In this paper, we introduce the concept of wavelet-based contourlet packets (WBCP), which is similar to the notion of wavelet packets (WP). Using WBCP, we have the flexibility of choosing the most proper basis based on a criterion. In this work, we utilize WBCP for image coding to extend our previous work that was based on WBCT for image coding. Our simulation results show that the proposed WBCT packets provide both visual and PSNR improvements over WBCT. Moreover, for texture images the results outperform those of WP, visually, while achieve comparable PSNR values.
The area of domain adaptation has been instrumental in addressing the domain shift problem encoun... more The area of domain adaptation has been instrumental in addressing the domain shift problem encountered by many applications. This problem arises due to the difference between the distributions of source data used for training in comparison with target data used during realistic testing scenarios. In this paper, we introduce a novel MultiScale Domain Adaptive YOLO (MS-DAYOLO) framework that employs multiple domain adaptation paths and corresponding domain classifiers at different scales of the recently introduced YOLOv4 object detector to generate domain-invariant features. We train and test our proposed method using popular datasets. Our experiments show significant improvements in object detection performance when training YOLOv4 using the proposed MS-DAYOLO and when tested on target data representing challenging weather conditions for autonomous driving applications.
In this paper, we analyze statistical and rate-distortion (R-D) properties of MPEG-4 Fine-Granula... more In this paper, we analyze statistical and rate-distortion (R-D) properties of MPEG-4 Fine-Granular Scalability (FGS), which has recently become an important scalable compression framework and a de-facto standard for Internet video streaming. We first propose a novel statistical model of DCT residue that accurately captures the properties of the input to the MPEG-4 FGS enhancement layer. Our results show that FGS residue concentrates a lot of probability mass near zero and cannot be accurately modeled by Gaussian or Laplacian distributions. We then model the distortion of each bitplane based on the proposed statistical framework and further demonstrate that our R-D model significantly outperforms current distortion models.
The convergence of the Internet with new wireless and mobile networks is creating a whole new lev... more The convergence of the Internet with new wireless and mobile networks is creating a whole new level of heterogeneity in multimedia communications. This increased level of heterogeneity emphasizes the need for scalable and adaptive video solutions both for coding and transmission purposes. However, in general, there is an inherent tradeoff between the level of scalability and the quality of scalable video streams. In other words, the higher the bandwidth variation, the lower the overall video quality of the scalable stream that is needed to support the desired bandwidth range. In this paper, we introduce the notion of TranScaling (TS) which is a generalization of (non-scalable) transcoding. With transcaling, a scalable video stream, that covers a given bandwidth range, is mapped into one or more scalable video streams covering different bandwidth ranges. Our proposed TS framework exploits the fact that the level of heterogeneity changes at different points of the video distribution tree over wireless and mobile Internet networks. This provides the opportunity to improve the video quality by performing the appropriate transcaling process. We argue that an Internet/wireless network gateway represents a good candidate for performing transcaling. Moreover, we describe Hierarchical TranScaling (HTS) which provides a "Transcalar" the option of choosing among different levels of transcaling processes with different complexities. We illustrate the benefits of transcaling by considering the recently developed MPEG-4 Fine-Granular-Scalability (FGS) video coding. Simulation results of video transcaling are also presented.
Radar Tracking With Orthogonal Velocity Measurements for Autonomous Ground Vehicles
In this work we examine the effect of having independent radial and angular velocity radar measur... more In this work we examine the effect of having independent radial and angular velocity radar measurements on 2D tracking performance. With this, we provide design considerations for systems using interferometric angular velocity estimation. We found that on average as the number of clutter detections increases, the reduction in error by adding an additional angular velocity measurement also increases, implying a greater need for orthogonal velocity measurements in high-clutter environments. In particular, in simulation at a clutter rate of 50 detections per measurement, we obtained an error reduction of 14% by adding a radial velocity measurement and a further reduction of 3% from an additional angular velocity measurement. We also tested this method on a publicly available autonomous driving dataset by synthesizing angular velocity measurements and found a similar reduction of 13% and 6% by adding Doppler and angular velocity measurements, respectively.
Multimedia Over Wireless
CRC Press eBooks, Mar 22, 2000
Notice of Removal Hyperspectral material classification under monochromatic and trichromatic sampling rates
3D Multi-Object Tracking using Random Finite Set-based Multiple Measurement Models Filtering (RFS-M3) for Autonomous Vehicles
Multiple object tracking (MOT) is a critical module for enabling autonomous vehicles to achieve s... more Multiple object tracking (MOT) is a critical module for enabling autonomous vehicles to achieve safe planing and navigation in cluttered environments. In tracking-by-detection systems, there are inevitably many false positives and misses among learning-based input detections. The challenge for MOT is to combine these detections into tracks, and filter them based on their uncertainties, states, and temporal consistency to achieve accurate and persistent tracks. In this paper, we propose to solve the 3D MOT problem for autonomous driving applications using a random finite set-based (RFS) Multiple Measurement Models filter (RFS-M3). In partiuclar, we propose multiple measurement models for a Poisson multi-Bernoulli mixture (PMBM) filter in support of different application scenarios. Our RFS-M3 filter can naturally model these uncertainties accurately and elegantly. We combine the learning-based detections with our RFS-M3 tracker through incorporating the detection confidence score into the PMBM prediction and update step. The superior experimental results of our RFS-M3 tracker on Waymo, Argoverse and nuSceness datasets illustrate that our RFS-M3 tracker outperforms state-of-the-art deep learning-based and traditional filter-based approaches. To the best of our knowledge, this represents a first successful attempt for employing an RFS-based approach in conjunction with 3D learning-based amodal detections for 3D MOT applications with comprehensive validation using challenging datasets made available by industry leaders.
Under traditional IP multicast, application-level FEC can only be implemented on an end-to-end ba... more Under traditional IP multicast, application-level FEC can only be implemented on an end-to-end basis between the sender and the clients. Emerging overlay and peer-to-peer (p2p) networks open the door for new paradigms of network FEC. The deployment of FEC within these emerging networks has received very little attention (if any). In this paper, we analyze and optimize the impact of Network-Embedded FEC (NEF) in overlay and p2p multimedia multicast networks. Under NEF, we place FEC codecs in selected intermediate nodes of a multicast tree. The NEF codecs detect and recover lost packets within FEC blocks at earlier stages before these blocks arrive at deeper intermediate nodes or at the final leaf nodes. This approach significantly reduces the probability of receiving undecodable FEC blocks. In essence, the proposed NEF codecs work as signal regenerators in a communication system and can reconstruct most of the lost data packets without requiring retransmission. We develop an optimization algorithm for the placement of NEF codecs within random multicast trees. Our theoretical analysis and simulation results show that a relatively small number of NEF codecs placed in (sub-)optimally selected intermediate nodes of a network can improve the throughput and overall reliability dramatically.
arXiv (Cornell University), Sep 1, 2020
There have been significant advances in neural networks for both 3D object detection using LiDAR ... more There have been significant advances in neural networks for both 3D object detection using LiDAR and 2D object detection using video. However, it has been surprisingly difficult to train networks to effectively use both modalities in a way that demonstrates gain over single-modality networks. In this paper, we propose a novel Camera-LiDAR Object Candidates (CLOCs) fusion network. CLOCs fusion provides a low-complexity multi-modal fusion framework that significantly improves the performance of single-modality detectors. CLOCs operates on the combined output candidates before Non-Maximum Suppression (NMS) of any 2D and any 3D detector, and is trained to leverage their geometric and semantic consistencies to produce more accurate final 3D and 2D detection results. Our experimental evaluation on the challenging KITTI object detection benchmark, including 3D and bird's eye view metrics, shows significant improvements, especially at long distance, over the state-of-the-art fusion based methods. At time of submission, CLOCs ranks the highest among all the fusion-based methods in the official KITTI leaderboard. We will release our code upon acceptance.
Integrated Generative-Model Domain-Adaptation for Object Detection under Challenging Conditions
2022 IEEE 95th Vehicular Technology Conference: (VTC2022-Spring), Jun 1, 2022
Poster Abstract: Wind Speed and Direction Estimation Using Manifold Approximation
ABSTRACT
Analysis and design of reliable and stable link-layer protocols for wireless communication
ABSTRACT
Rain-Adaptive Intensity-Driven Object Detection for Autonomous Vehicles
SAE technical paper series, Apr 14, 2020
arXiv (Cornell University), May 7, 2014
arXiv (Cornell University), Jun 30, 2020
Advanced automotive active-safety systems, in general, and autonomous vehicles, in particular, re... more Advanced automotive active-safety systems, in general, and autonomous vehicles, in particular, rely heavily on visual data to classify and localize objects such as pedestrians, traffic signs and lights, and other nearby cars, to assist the corresponding vehicles maneuver safely in their environments. However, the performance of object detection methods could degrade rather significantly under challenging weather scenarios including rainy conditions. Despite major advancements in the development of deraining approaches, the impact of rain on object detection has largely been understudied, especially in the context of autonomous driving. The main objective of this paper is to present a tutorial on state-of-the-art and emerging techniques that represent leading candidates for mitigating the influence of rainy conditions on an autonomous vehicle's ability to detect objects. Our goal includes surveying and analyzing the performance of object detection methods trained and tested using visual data captured under clear and rainy conditions. Moreover, we survey and evaluate the efficacy and limitations of leading deraining approaches, deep-learning based domain adaptation, and image translation frameworks that are being considered for addressing the problem of object detection under rainy conditions. Experimental results of a variety of the surveyed techniques are presented as part of this tutorial.
PEEC: A Channel-Adaptive Feedback-Based Error Control Protocol for Wireless MAC Layer
Abstract—Reliable transmission is a challenging task over wireless LANs since wireless links are ... more Abstract—Reliable transmission is a challenging task over wireless LANs since wireless links are known to be susceptible to errors. Although the current IEEE802.11 standard ARQ error control protocol performs relatively well over channels with very low bit error rates (BERs), this performance deteriorates rapidly as the BER increases. This paper investigates the problem of reliable transmission in a contention free wireless LAN and introduces a Packet Embedded Error Control (PEEC) protocol, which employs packet-embedded parity symbols instead of ARQbased retransmission for error recovery. Specifically, depending on receiver feedback, PEEC adaptively estimates channel conditions and administers the transmission of (data and parity) symbols within a packet. This enables successful recovery of both new data and old unrecovered data from prior transmissions. In addition to theoretically analyzing PEEC, the performance of the proposed scheme is extensively analyzed over real channel traces collected on 802.11b WLANs. We compare PEEC performance with the performance of the IEEE802.11 standard ARQ protocol as well as contemporary protocols such as enhanced ARQ and the hybrid ARQ/FEC. Our analysis and experimental simulations show that PEEC outperforms all three competing protocols over a wide range of actual 802.11b WLAN collected traces. Finally, the design and implementation of PEEC using an Adaptive Low-Density-Parity-Check (A-LDPC) decoder is presented. Index Terms—Forward error correction, Channel coding, Wireless LAN, Feedback communication, Automatic repeat request.
Real-time switching of MPEG-2 bitstreams
ABSTRACT
Fast image super-resolution via selective manifold learning of high-resolution patches
This paper considers the problem of single image super-resolution (SR). Previous example-based SR... more This paper considers the problem of single image super-resolution (SR). Previous example-based SR approaches mainly focus on analyzing the co-occurrence properties of low resolution (LR) and high resolution (HR) patches via dictionary learning. In our recent work [1], a novel approach (SR via sparse subspace clustering-based linear approximation of manifold or SLAM) has been proposed. In this paper, we further improve the SLAM method by considering and analyzing each tangent subspace as one point in a Grassmann manifold to select an optimal subset of tangent spaces. Furthermore, the optimal subset is clustered hierarchically, which helps in reducing the proposed algorithm's complexity significantly while still preserving the quality of the reconstructed HR image.
The contourlet transform is a new directional transform, which is capable of capturing contours a... more The contourlet transform is a new directional transform, which is capable of capturing contours and fine details in images. We recently introduced the wavelet-based contourlet transform (WBCT) that is a non-redundant version of the contourlet transform, and appropriately used this transform for image coding. In this paper, we introduce the concept of wavelet-based contourlet packets (WBCP), which is similar to the notion of wavelet packets (WP). Using WBCP, we have the flexibility of choosing the most proper basis based on a criterion. In this work, we utilize WBCP for image coding to extend our previous work that was based on WBCT for image coding. Our simulation results show that the proposed WBCT packets provide both visual and PSNR improvements over WBCT. Moreover, for texture images the results outperform those of WP, visually, while achieve comparable PSNR values.