Alert button
Picture for Siniša Šegvić

Siniša Šegvić

Alert button

Outlier detection by ensembling uncertainty with negative objectness

Feb 23, 2024
Anja Delić, Matej Grcić, Siniša Šegvić

Viaarxiv icon

Real time dense anomaly detection by learning on synthetic negative data

May 24, 2023
Anja Delić, Matej Grcić, Siniša Šegvić

Figure 1 for Real time dense anomaly detection by learning on synthetic negative data
Figure 2 for Real time dense anomaly detection by learning on synthetic negative data
Viaarxiv icon

Identifying Label Errors in Object Detection Datasets by Loss Inspection

Mar 13, 2023
Marius Schubert, Tobias Riedlinger, Karsten Kahl, Daniel Kröll, Sebastian Schoenen, Siniša Šegvić, Matthias Rottmann

Figure 1 for Identifying Label Errors in Object Detection Datasets by Loss Inspection
Figure 2 for Identifying Label Errors in Object Detection Datasets by Loss Inspection
Figure 3 for Identifying Label Errors in Object Detection Datasets by Loss Inspection
Figure 4 for Identifying Label Errors in Object Detection Datasets by Loss Inspection
Viaarxiv icon

Normalizing Flow based Feature Synthesis for Outlier-Aware Object Detection

Feb 01, 2023
Nishant Kumar, Siniša Šegvić, Abouzar Eslami, Stefan Gumhold

Figure 1 for Normalizing Flow based Feature Synthesis for Outlier-Aware Object Detection
Figure 2 for Normalizing Flow based Feature Synthesis for Outlier-Aware Object Detection
Figure 3 for Normalizing Flow based Feature Synthesis for Outlier-Aware Object Detection
Figure 4 for Normalizing Flow based Feature Synthesis for Outlier-Aware Object Detection
Viaarxiv icon

Hybrid Open-set Segmentation with Synthetic Negative Data

Jan 19, 2023
Matej Grcić, Siniša Šegvić

Figure 1 for Hybrid Open-set Segmentation with Synthetic Negative Data
Figure 2 for Hybrid Open-set Segmentation with Synthetic Negative Data
Figure 3 for Hybrid Open-set Segmentation with Synthetic Negative Data
Figure 4 for Hybrid Open-set Segmentation with Synthetic Negative Data
Viaarxiv icon

On advantages of Mask-level Recognition for Open-set Segmentation in the Wild

Jan 09, 2023
Matej Grcić, Josip Šarić, Siniša Šegvić

Figure 1 for On advantages of Mask-level Recognition for Open-set Segmentation in the Wild
Figure 2 for On advantages of Mask-level Recognition for Open-set Segmentation in the Wild
Figure 3 for On advantages of Mask-level Recognition for Open-set Segmentation in the Wild
Figure 4 for On advantages of Mask-level Recognition for Open-set Segmentation in the Wild
Viaarxiv icon

Weakly supervised training of universal visual concepts for multi-domain semantic segmentation

Dec 20, 2022
Petra Bevandić, Marin Oršić, Ivan Grubišić, Josip Šarić, Siniša Šegvić

Figure 1 for Weakly supervised training of universal visual concepts for multi-domain semantic segmentation
Figure 2 for Weakly supervised training of universal visual concepts for multi-domain semantic segmentation
Figure 3 for Weakly supervised training of universal visual concepts for multi-domain semantic segmentation
Figure 4 for Weakly supervised training of universal visual concepts for multi-domain semantic segmentation
Viaarxiv icon

Dynamic loss balancing and sequential enhancement for road-safety assessment and traffic scene classification

Nov 08, 2022
Marin Kačan, Marko Ševrović, Siniša Šegvić

Figure 1 for Dynamic loss balancing and sequential enhancement for road-safety assessment and traffic scene classification
Figure 2 for Dynamic loss balancing and sequential enhancement for road-safety assessment and traffic scene classification
Figure 3 for Dynamic loss balancing and sequential enhancement for road-safety assessment and traffic scene classification
Figure 4 for Dynamic loss balancing and sequential enhancement for road-safety assessment and traffic scene classification
Viaarxiv icon

Automatic universal taxonomies for multi-domain semantic segmentation

Jul 18, 2022
Petra Bevandić, Siniša Šegvić

Figure 1 for Automatic universal taxonomies for multi-domain semantic segmentation
Figure 2 for Automatic universal taxonomies for multi-domain semantic segmentation
Figure 3 for Automatic universal taxonomies for multi-domain semantic segmentation
Figure 4 for Automatic universal taxonomies for multi-domain semantic segmentation
Viaarxiv icon

DenseHybrid: Hybrid Anomaly Detection for Dense Open-set Recognition

Jul 06, 2022
Matej Grcić, Petra Bevandić, Siniša Šegvić

Figure 1 for DenseHybrid: Hybrid Anomaly Detection for Dense Open-set Recognition
Figure 2 for DenseHybrid: Hybrid Anomaly Detection for Dense Open-set Recognition
Figure 3 for DenseHybrid: Hybrid Anomaly Detection for Dense Open-set Recognition
Figure 4 for DenseHybrid: Hybrid Anomaly Detection for Dense Open-set Recognition
Viaarxiv icon