DriveGazen: Event-Based Driving Status Recognition Using Conventional Camera

Authors: Xiaoyin Yang, Xin Yang

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We specifically collected the Driving Status (Drive Gaze) dataset to demonstrate the effectiveness of our approach. Additionally, we validate the superiority of the Drive Gazen on the Single-eye Event-based Emotion (SEE) dataset. To the best of our knowledge, our method is the first to utilize guide attention spiking neural networks and eye-based event frames generated from conventional cameras for driving status recognition.
Researcher Affiliation Academia Xiaoyin Yang, Xin Yang Dalian University of Technology EMAIL, EMAIL
Pseudocode No The paper describes the model architecture and processes using mathematical equations and descriptive text, but it does not include a clearly labeled pseudocode or algorithm block.
Open Source Code No The paper states: "Please refer to our project page and supplementary materials for more details." However, it does not provide a direct link to a code repository or an explicit statement confirming the immediate release of the source code for the described methodology.
Open Datasets Yes We specifically collected the Driving Status (Drive Gaze) dataset to demonstrate the effectiveness of our approach. Additionally, we validate the superiority of the Drive Gazen on the Single-eye Event-based Emotion (SEE) dataset. ... The first publicly available eye-based event-driven driving state dataset generated from conventional cameras, containing intensity frames and corresponding events, capturing data from different ages, races, genders, etc;
Dataset Splits Yes In total, Drive Gaze includes 1645 sequences/245365 frames of original events, with a total duration of 68.1 minutes(Figure 4, divided into 1316 for training and 329 for testing.
Hardware Specification Yes We trained ADSN for 150 epochs using a batch size of 128 on an NVIDIA TITAN V GPU.
Software Dependencies No ADSN is implemented in Py Torch (Paszke et al. 2019). While PyTorch is mentioned, a specific version number is not provided, nor are versions for any other key software dependencies.
Experiment Setup Yes We trained ADSN for 150 epochs using a batch size of 128 on an NVIDIA TITAN V GPU.For the SNN settings, we use a spiking threshold of 0.3 and a decay factor of 0.2 for all SNN neurons.