Summary of Enhanced Parking Perception by Multi-task Fisheye Cross-view Transformers, By Antonyo Musabini et al.
Enhanced Parking Perception by Multi-Task Fisheye Cross-view Transformers
by Antonyo Musabini, Ivan Novikov, Sana Soula, Christel Leonet, Lihao Wang, Rachid Benmokhtar, Fabian Burger, Thomas Boulay, Xavier Perrotton
First submitted to arxiv on: 22 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper introduces Multi-Task Fisheye Cross View Transformers (MT F-CVT), a novel algorithm for parking area perception. Unlike previous approaches, which rely on homographic projection and are limited to detecting vacant slots within a range, MT F-CVT leverages features from a four-camera fisheye Surround-view Camera System (SVCS) with multihead attentions to create a detailed Bird-Eye View (BEV) grid feature map. The algorithm processes features using both a segmentation decoder and a Polygon-Yolo based object detection decoder for parking slots and vehicles. Trained on LiDAR-labeled data, MT F-CVT achieves an average error of 20 cm in real open-road scenes up to 25m x 25m. The larger model achieves an F-1 score of 0.89, while the smaller model operates at 16 fps on an Nvidia Jetson Orin embedded board with similar detection results. MT F-CVT demonstrates robust generalization capability across different vehicles and camera rig configurations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about creating a better way to understand what’s happening in parking areas using cameras. Right now, most algorithms can only see a small part of the parking area and are not very good at detecting things like where cars are parked or which spaces are available. The new algorithm, called MT F-CVT, uses information from multiple cameras to create a big picture view of the parking area. This allows it to detect things like car positions and availability with high accuracy. The algorithm is also fast enough to be used in real-time applications like self-driving cars. |
Keywords
* Artificial intelligence * Decoder * Feature map * Generalization * Multi task * Object detection * Yolo
