Structured Output Prediction for Semantic Perception in Autonomous Vehicles
Authors: Rein Houthooft, Cedric De Boom, Stijn Verstichel, Femke Ongenae, Filip De Turck
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Both quantitative and qualitative analyses demonstrate that by taking into account contextual relations between pixel segmentation regions within a second-degree neighborhood, spurious label assignments are filtered out, leading to highly accurate semantic segmentations for outdoor scenes. |
| Researcher Affiliation | Academia | Department of Information Technology (INTEC) Ghent University i Minds, Gaston Crommenlaan 8 box 201, B-9050 Ghent, Belgium {rein.houthooft, cedric.deboom, stijn.verstichel, femke.ongenae, filip.deturck}@ugent.be |
| Pseudocode | No | The paper describes algorithms and models mathematically and textually but does not include any pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any statement about releasing source code or a link to a code repository for the methodology described. |
| Open Datasets | Yes | Furthermore, we add quantitative performance results based on two publicly available datasets to objectively evaluate our model. Specifically, the autonomous vehicle segmentation dataset KITTI (Ros et al. 2015) is used, augmented with 49 additional train images by (Kundu et al. 2014), leading to 149 train and 46 test images, and the Cam Vid dataset (Brostow, Fauqueur, and Cipolla 2009), split into 525 train and 175 test images. |
| Dataset Splits | Yes | Model hyperparmeters are optimized via k-fold crossvalidation on the training set. ...Specifically, the autonomous vehicle segmentation dataset KITTI (Ros et al. 2015) is used, augmented with 49 additional train images by (Kundu et al. 2014), leading to 149 train and 46 test images, and the Cam Vid dataset (Brostow, Fauqueur, and Cipolla 2009), split into 525 train and 175 test images. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running experiments (e.g., CPU/GPU models, memory specifications). |
| Software Dependencies | No | The paper mentions several software components and algorithms like SLIC, DAISY, SSVM, and α-expansion, but it does not specify any version numbers for these or other software dependencies (e.g., Python, PyTorch, TensorFlow). |
| Experiment Setup | Yes | Model hyperparmeters are optimized via k-fold crossvalidation on the training set. ...Input images are first segmented into 1000 regions with a SLIC compactness of 15. The gradient and the color bag-of-words vectors dimensions are set to G = 300 and C = 150 (further increment of these dimensions did not lead to additional performance gains). |