Deep Event Stereo Leveraged by Event-to-Image Translation

Authors: Soikat Hasan Ahmed, Hae Woong Jang, S M Nadim Uddin, Yong Ju Jung882-890

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results reveal that our method outperforms the state-of-the-art methods by significant margins both in quantitative and qualitative measures.
Researcher Affiliation Academia College of Information Technology Convergence, Gachon University, Seongnam, South Korea
Pseudocode No The paper describes the architecture and various sub-networks with detailed textual explanations and figures, but it does not include any explicit pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any explicit statement about releasing source code for the described methodology or a link to a code repository.
Open Datasets Yes We evaluate our proposed method on the Multi Vehicle Stereo Event Camera Dataset (MVSEC) (Zhu et al. 2018a).
Dataset Splits Yes In the split one, we train the model using 3110 samples from the Indoor Flying 2-3 and for the validation and test, we use 200 and 861 samples from the Indoor Flying 1 sequence, respectively. In the split three, we train the model with 2600 samples from the Indoor Flying 1-2 and for the validation and test, we use 200 and 1343 samples from the Indoor Flying 3, respectively.
Hardware Specification Yes A single NVIDIA TITAN XP GPU was used for the training.
Software Dependencies No The proposed deep event stereo network was implemented using Py Torch.
Experiment Setup Yes The model was trained in an end-to-end manner with the RMSprop optimizer using default settings. We trained the model up to 15 epoch and chose the best checkpoint based on the validation results for the testing.