Summary of Alpi: Auto-labeller with Proxy Injection For 3d Object Detection Using 2d Labels Only, by Saad Lahlali et al.
ALPI: Auto-Labeller with Proxy Injection for 3D Object Detection using 2D Labels Only
by Saad Lahlali, Nicolas Granger, Hervé Le Borgne, Quoc-Cuong Pham
First submitted to arxiv on: 24 Jul 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 The proposed weakly supervised 3D annotator uses solely 2D bounding box annotations from images, along with size priors to train a 3D detector. The method introduces a simple yet effective solution by constructing 3D proxy objects and adding them to the training dataset. To align 2D supervision with 3D detection, the method ensures depth invariance through a novel expression of 2D losses. Additionally, an offline pseudo-labelling scheme is used to gradually improve 3D pseudo-labels. The method achieves performance comparable or above previous works on the Car category and close to fully supervised methods on more challenging classes. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A team of researchers developed a new way to detect objects in 3D space using only 2D information from images. This is helpful because it makes it easier to label data for machines that can see in 3D, like self-driving cars. The method uses a combination of existing ideas and some new tricks to get the job done. It works by creating fake 3D objects based on what we already know about 2D objects, then using those fake objects to help train the machine learning model. |
Keywords
» Artificial intelligence » Bounding box » Machine learning » Supervised
