|
Assessment of Uncertainty of Neural Net Predictions in the Tasks of Classification, Detection and Segmentation
Vlasák, Jiří ; Kohút, Jan (referee) ; Herout, Adam (advisor)
This work focuses on comparing three widely used methods for improving uncertainty estimations: Deep Ensembles, Monte Carlo Dropout, and Temperature Scaling. These methods are applied to six computer vision models that are pretrained as well as trained from scratch. The models are then evaluated on computer vision datasets for classification, semantic segmentation, and object detection using a wide range of metrics. The models are also evaluated on distorted versions of these datasets to measure their performance on out-of-distribution data. These modified models achieve promising results. Ensembles outperform the other models by as high as 70 % in accuracy and 0.2 in IOU on the distorted MedSeg COVID-19 segmentation dataset while also outperforming the other models on the CIFAR-100 and FMNIST datasets.
|
| |
| |
| |
|
Variant design of transport connections of shopping centre in Modřice
Knopp, Martin ; Fencl, Ivan (referee) ; Holcner, Petr (advisor)
This diploma thesis deals with a proposal and a comparison of several options of transportation planning in the shopping area in the urban area of the town of Modřice, which is enlarging due to the building-up of new commercial centres. The first part of the thesis consists of a treatise on the field of traffic engineering, road-traffic telematics and ITS. The core of the work is formed by a detailed analysis of the current state and calibration of the transport simulation system AIMSUN. The following part shows three different ways of transportation in the given area and all of them include a model of the particular area made in the program AIMSUN and an elaborate analysis of the proposed solution. In the last part of the work all the proposed solutions are compared according to the results gained from the AIMSUN model.
|
|
Assessment of Uncertainty of Neural Net Predictions in the Tasks of Classification, Detection and Segmentation
Vlasák, Jiří ; Kohút, Jan (referee) ; Herout, Adam (advisor)
This work focuses on comparing three widely used methods for improving uncertainty estimations: Deep Ensembles, Monte Carlo Dropout, and Temperature Scaling. These methods are applied to six computer vision models that are pretrained as well as trained from scratch. The models are then evaluated on computer vision datasets for classification, semantic segmentation, and object detection using a wide range of metrics. The models are also evaluated on distorted versions of these datasets to measure their performance on out-of-distribution data. These modified models achieve promising results. Ensembles outperform the other models by as high as 70 % in accuracy and 0.2 in IOU on the distorted MedSeg COVID-19 segmentation dataset while also outperforming the other models on the CIFAR-100 and FMNIST datasets.
|
| |
| |
| |
|
Variant design of transport connections of shopping centre in Modřice
Knopp, Martin ; Fencl, Ivan (referee) ; Holcner, Petr (advisor)
This diploma thesis deals with a proposal and a comparison of several options of transportation planning in the shopping area in the urban area of the town of Modřice, which is enlarging due to the building-up of new commercial centres. The first part of the thesis consists of a treatise on the field of traffic engineering, road-traffic telematics and ITS. The core of the work is formed by a detailed analysis of the current state and calibration of the transport simulation system AIMSUN. The following part shows three different ways of transportation in the given area and all of them include a model of the particular area made in the program AIMSUN and an elaborate analysis of the proposed solution. In the last part of the work all the proposed solutions are compared according to the results gained from the AIMSUN model.
|