Car cabin object detection using artificial intelligence (multimodal object detection)
Artificial intelligence has advanced tremendously in recent years, notably in areas such as computer vision. Object detection is a useful technique for tracking driver movements and improving road safety. Cameras can identify a wide range of objects, including people, computers, phones, infants, and...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/176972 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Artificial intelligence has advanced tremendously in recent years, notably in areas such as computer vision. Object detection is a useful technique for tracking driver movements and improving road safety. Cameras can identify a wide range of objects, including people, computers, phones, infants, and many more, allowing us to alert drivers to potential safety hazards when they start to get fatigued or distracted.
Red Green Blue (RGB) cameras have long been the industry standard for a wide range of computer vision applications. However, in a low lighting environment, they usually struggle to obtain accurate readings. This limitation reduces the efficiency of object detecting systems and constitute a severe concern, especially if there are false alarms or missed detections.
In this paper, we study the integration of RGB and infrared (IR) channels to leverage on the strength of each modality under different lighting condition. Literature review was done on the state-of-art methods for multi-modality object detection, and an adaptive dual-discrepancy calibration network is proposed to tackle the misalignment issue when fusing the two modalities. |
|---|