4,171 Hits in 2.8 sec

Deep Deterministic Uncertainty for Semantic Segmentation [article]

Jishnu Mukhoti, Joost van Amersfoort, Philip H.S. Torr, Yarin Gal
2021 arXiv   pre-print
We extend Deep Deterministic Uncertainty (DDU), a method for uncertainty estimation using feature space densities, to semantic segmentation.  ...  Using the DeepLab-v3+ architecture on Pascal VOC 2012, we show that DDU improves upon MC Dropout and Deep Ensembles while being significantly faster to compute.  ...  Conclusion In this paper, we show that Deep Deterministic Uncertainty (DDU) can be easily extended to the task of semantic segmentation.  ... 
arXiv:2111.00079v1 fatcat:xal74vuiwfdongiio2d3dy2p3e

On the Practicality of Deterministic Epistemic Uncertainty [article]

Janis Postels, Mattia Segu, Tao Sun, Luc Van Gool, Fisher Yu, Federico Tombari
2021 arXiv   pre-print
Then, we extend the most promising approaches to semantic segmentation.  ...  A set of novel approaches for estimating epistemic uncertainty in deep neural networks with a single forward pass has recently emerged as a valid alternative to Bayesian Neural Networks.  ...  In the dilated resnet architecture used for semantic segmentation the latent representation z is passed through a point-wise feedforward layer f : R Wz×Hz×Cz → R Wz×Hz×3 and, subsequently, bilinearly upsampled  ... 
arXiv:2107.00649v2 fatcat:hpo6z42vifej7lgdmx4tje6scu

Large-Scale Visual Active Learning with Deep Probabilistic Ensembles [article]

Kashyap Chitta, Jose M. Alvarez, Adam Lesnikowski
2019 arXiv   pre-print
Annotating the right data for training deep neural networks is an important challenge.  ...  We conduct a series of large-scale visual active learning experiments to evaluate DPEs on classification with the CIFAR-10, CIFAR-100 and ImageNet datasets, and semantic segmentation with the BDD100k dataset  ...  Active Semantic Segmentation The goal of semantic segmentation is to assign a class label to every pixel in an image.  ... 
arXiv:1811.03575v3 fatcat:5luahiq67zbv7ffvyqbauvm4uu

Bayesian deep learning of affordances from RGB images [article]

Lorenzo Mur-Labadia, Ruben Martinez-Cantin
2021 arXiv   pre-print
However, previous works assume a deterministic model, but uncertainty quantification is fundamental for robust detection, affordance-based reason, continual learning, etc.  ...  For comparison, we estimate the uncertainty using two state-of-the-art techniques: Monte Carlo dropout and deep ensembles. We also compare different types of CNN encoders for feature extraction.  ...  Kendall and Gal [10] analyzed both uncertainties in common computer vision tasks (semantic segmentation and depth regression) to show that the aleatoric and epistemic uncertainty model different phenomenons  ... 
arXiv:2109.12845v1 fatcat:n7gqlxomina75msjhret3gak7i

Predictive Monitoring with Logic-Calibrated Uncertainty for Cyber-Physical Systems [article]

Meiyi Ma, John Stankovic, Ezio Bartocci, Lu Feng
2021 arXiv   pre-print
We develop a novel approach for monitoring sequential predictions generated from Bayesian Recurrent Neural Networks (RNNs) that can capture the inherent uncertainty in CPS, drawing on insights from our  ...  We define STL-U strong and weak satisfaction semantics based on if all or some sequences contained in a flowpipe satisfy the requirement.  ...  Uncertainty Estimation with Bayesian RNN Models Stochastic regularization techniques (SRTs) have been popularly used to cast deterministic deep learning models as Bayesian models for uncertainty estimation  ... 
arXiv:2011.00384v3 fatcat:gxufwlzxnbgqvja7wwtkqtey5i

Bayesian Deep Learning for Segmentation for Autonomous Safe Planetary Landing [article]

Kento Tomita and Katherine A. Skinner and Koki Ho
2022 arXiv   pre-print
The developed approach enables reliable, safe landing site detection by: (i) generating simultaneously a safety prediction map and its uncertainty map via Bayesian deep learning and semantic segmentation  ...  In response to these limitations, this paper proposes an application of the Bayesian deep-learning segmentation method for hazard detection.  ...  Bayesian Deep Learning for Semantic Segmentation Figure 2 shows the network architecture used for the semantic segmentation stage, which is based on Bayesian SegNet [34] .  ... 
arXiv:2102.10545v2 fatcat:fbgg4zwn5jduvlxwsdlzbxyroa

Scalable Uncertainty for Computer Vision with Functional Variational Inference [article]

Eduardo D C Carvalho, Ronald Clark, Andrea Nicastro, Paul H J Kelly
2020 arXiv   pre-print
As Deep Learning continues to yield successful applications in Computer Vision, the ability to quantify all forms of uncertainty is a paramount requirement for its safe and reliable deployment in the real-world  ...  segmentation.  ...  We would like to thank Jan Czarnowski, Sajad Saeedi, Tristan Laidlow and all our reviewers for helpful insights and comments.  ... 
arXiv:2003.03396v1 fatcat:3nwrztbfhnf37ln27j6xrvgute

Evaluating Bayesian Deep Learning Methods for Semantic Segmentation [article]

Jishnu Mukhoti, Yarin Gal
2019 arXiv   pre-print
Deep learning has been revolutionary for computer vision and semantic segmentation in particular, with Bayesian Deep Learning (BDL) used to obtain uncertainty maps from deep models when predicting semantic  ...  This information is critical when using semantic segmentation for autonomous driving for example. Standard semantic segmentation systems have well-established evaluation metrics.  ...  There has been a lot of research on deep architectures for semantic segmentation [46, 55, 5, 9, 11, 10, 13, 66] .  ... 
arXiv:1811.12709v2 fatcat:lqtc6w3n6jehtcma4m3athavwu

Risk-Aware Planning by Confidence Estimation using Deep Learning-Based Perception [article]

Maymoonah Toubeh, Pratap Tokekar
2019 arXiv   pre-print
Experiments are conducted using a deep learning semantic image segmentation, followed by a path planner based on the resulting cost map, to provide an empirical analysis of the proposed method.  ...  This work proposes the use of Bayesian approximations of uncertainty from deep learning in a robot planner, showing that this produces more cautious actions in safety-critical scenarios.  ...  In another work, the uncertainty from dropout is used in semantic segmentation for improved learning and test time estimation [12] .  ... 
arXiv:1910.00101v1 fatcat:wzhsojsfbrh7vh54gbug3imfui

Scribble-Supervised Semantic Segmentation by Uncertainty Reduction on Neural Representation and Self-Supervision on Neural Eigenspace [article]

Zhiyi Pan, Peng Jiang, Yunhai Wang, Changhe Tu, Anthony G. Cohn
2021 arXiv   pre-print
Scribble-supervised semantic segmentation has gained much attention recently for its promising performance without high-quality annotations.  ...  Instead, this work aims to achieve semantic segmentation by scribble annotations directly without extra information and other limitations.  ...  To our best knowledge, minimizing entropy is adopted to scribble-supervised semantic segmentation for uncertainty reduction for the first time.  ... 
arXiv:2102.09896v1 fatcat:adxb7ie5jjgmvcvcc7yl6pw6om

Uncertainty Aware Proposal Segmentation for Unknown Object Detection [article]

Yimeng Li, Jana Kosecka
2021 arXiv   pre-print
semantic segmentation.  ...  We use object proposals generated by Region Proposal Network (RPN) and adapt distance aware uncertainty estimation of semantic segmentation using Radial Basis Functions Networks (RBFN) for class agnostic  ...  Introduction The last decade marked big progress in the design of Deep Network models for object detection and semantic segmentation.  ... 
arXiv:2111.12866v1 fatcat:bx4dolp2hvgmvbywyubj7dk65a

A Review on Deep Learning Techniques for Video Prediction [article]

Sergiu Oprea, Pablo Martinez-Gonzalez, Alberto Garcia-Garcia, John Alejandro Castro-Vargas, Sergio Orts-Escolano, Jose Garcia-Rodriguez, Antonis Argyros
2020 arXiv   pre-print
Motivated by the increasing interest in this task, we provide a review on the deep learning methods for prediction in video sequences.  ...  In light of the success of deep learning in computer vision, deep-learning-based video prediction emerged as a promising research direction.  ...  Despite its popularity, the original dataset did not contain ground truth for semantic segmentation.  ... 
arXiv:2004.05214v2 fatcat:weerbkanmjb4dn6wkn5o4b5aia

Risk-Aware Planning and Assignment for Ground Vehicles using Uncertain Perception from Aerial Vehicles [article]

Vishnu D. Sharma, Maymoonah Toubeh, Lifeng Zhou, Pratap Tokekar
2020 arXiv   pre-print
Deep learning techniques can be used for semantic segmentation of the aerial image to create a cost map for safe ground robot navigation. Such segmentation may still be noisy.  ...  Our contributions are two-fold: (i) we show how to use Bayesian deep learning techniques to extract risk at the perception level; and (ii) use a risk-theoretical measure, CVaR, for risk-aware planning  ...  We implement a deep learning technique for semantic segmentation of the overhead image. Due to the uncertainty from segmentation, the travel cost of the vehicle turns out to be a random variable.  ... 
arXiv:2003.11675v2 fatcat:w6rb76tm7ze33ds6alsbgn2tui

Bayesian SegNet: Model Uncertainty in Deep Convolutional Encoder-Decoder Architectures for Scene Understanding [article]

Alex Kendall and Vijay Badrinarayanan and Roberto Cipolla
2016 arXiv   pre-print
We present a deep learning framework for probabilistic pixel-wise semantic segmentation, which we term Bayesian SegNet.  ...  Semantic segmentation is an important tool for visual scene understanding and a meaningful measure of uncertainty is essential for decision making.  ...  Real Time Performance Conclusions We have presented Bayesian SegNet, the first probabilistic framework for semantic segmentation using deep learning, which outputs a measure of model uncertainty for  ... 
arXiv:1511.02680v2 fatcat:2rc2pxlzm5ddlkksy7cmyhfoam

Vector Quantized Bayesian Neural Network Inference for Data Streams [article]

Namuk Park, Taekyu Lee, Songkuk Kim
2021 arXiv   pre-print
Experiments including semantic segmentation on real-world data show that this model performs significantly faster than BNNs while estimating predictive results comparable to or superior to the results  ...  Bayesian neural networks (BNN) can estimate the uncertainty in predictions, as opposed to non-Bayesian neural networks (NNs).  ...  Semantic Segmentation Semantic segmentation experiment, which is a pixel-wise classification, evaluates the computational and predictive performance of VQ-BNN with a modern deep NN in practical situation  ... 
arXiv:1907.05911v3 fatcat:zn6zjnpbubdfpi2rw3zsbe22oi
« Previous Showing results 1 — 15 out of 4,171 results