VarifocalNet: An IoU-aware Dense Object Detector
- Haoyang Zhang, Ying Wang, Feras Dayoub, N. Sunderhauf
- Computer ScienceComputer Vision and Pattern Recognition
- 31 August 2020
This paper proposes to learn an Iou-Aware Classification Score (IACS) as a joint representation of object presence confidence and localization accuracy and builds an IoU-aware dense object detector based on the FCOS+ATSS architecture, that they are called VarifocalNet or VFNet for short.
Using the Unscented Kalman Filter in Mono-SLAM with Inverse Depth Parametrization for Autonomous Airship Control
- N. Sunderhauf, S. Lange, P. Protzel
- MathematicsIEEE International Workshop on Safety, Security…
- 12 November 2007
It is shown how the Unscented Kalman Filter can be applied in a SLAM context with monocular vision and the recently published Inverse Depth Parametrization is used for undelayed single-hypothesis landmark initialization and modelling.
SWA Object Detection
This technique report systematically investigate the effects of applying SWA to object detection as well as instance segmentation and finds a good policy of performing SWA in object detection, and consistently achieves ∼1.0 AP improvement over various popular detectors on the challenging COCO benchmark.
Bayesian Controller Fusion: Leveraging Control Priors in Deep Reinforcement Learning for Robotics
- Krishan Rana, Vibhavari Dasagi, Jesse Haviland, Ben Talbot, Michael Milford, N. Sunderhauf
- Computer ScienceArXiv
- 21 July 2021
Bayesian Controller Fusion is a promising approach for combining the complementary strengths of deep RL and traditional robotic control, surpassing what either can achieve independently.
Semantics for Robotic Mapping, Perception and Interaction: A Survey
A taxonomy for semantics research in or relevant to robotics is established, split into four broad categories of activity, in which semantics are extracted, used, or both, and dozens of major topics including fundamentals from the computer vision field and key robotics research areas utilizing semantics are surveyed.
BenchBot: Evaluating Robotics Research in Photorealistic 3D Simulation and on Real Robots
The research benefits of using the BenchBot system are described, including: enhanced capacity to focus solely on research problems, direct quantitative feedback to inform research development, tools for deriving comprehensive performance characteristics, and submission formats which promote sharability and repeatability of research outcomes.
Online Monitoring of Object Detection Performance Post-Deployment
This work introduces a cascaded neural network that monitors the performance of the object detector by predicting the quality of its mean average precision (mAP) on a sliding window of the input frames.
Multiplicative Controller Fusion: A Hybrid Navigation Strategy For Deployment in Unknown Environments
- Krishan Rana, Vibhavari Dasagi, Ben Talbot, Michael Milford, N. Sunderhauf
- Computer ScienceArXiv
- 11 March 2020
This work presents a novel approach to model-free reinforcement learning that can leverage existing sub-optimal solutions as an algorithmic prior during training and deployment and shows the efficacy of the Multiplicative Controller Fusion approach on the task of robot navigation.
Evaluating the Impact of Semantic Segmentation and Pose Estimation on Dense Semantic SLAM
- S. Bista, David Hall, Ben Talbot, Haoyang Zhang, Feras Dayoub, N. Sunderhauf
- Computer ScienceIEEE/RJS International Conference on Intelligent…
- 16 September 2021
The quality of semantic maps generated by state-of-the-art class-and instance-aware dense semantic SLAM algorithms whose codes are publicly available are evaluated and the impacts both semantic segmentation and pose estimation have on the quality of semantics maps are explored.
Uncertainty for Identifying Open-Set Errors in Visual Object Detection
- Dimity Miller, N. Sunderhauf, Michael Milford, Feras Dayoub
- Computer ScienceIEEE Robotics and Automation Letters
- 3 April 2021
The results show that GMM-Det consistently outperforms existing uncertainty techniques for identifying and rejecting open-set detections, especially at the low-error-rate operating point required for safety-critical applications.