Zero-shot hazard identification in Autonomous Driving: A Case Study on the COOOL Benchmark
Lukas Picek, Vojtěch Čermák, Marek Hanzl · 2024
Read Zero-shot hazard identification in Autonomous Driving: A Case Study on the COOOL Benchmark by Lukas Picek, Vojtěch Čermák, Marek Hanzl (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-19944.txt.
Description
Lukas Picek, Vojtěch Čermák, and Marek Hanzl describe their entry to the COOOL autonomous driving competition in this two thousand twenty-four paper. The benchmark asks systems to detect and classify hazards that fall outside the usual training labels. Their pipeline handles three tasks: detecting the driver's reaction, identifying the hazard object, and writing a caption for it. Kernel-based change point detection on bounding boxes and optical flow drives the reaction task.
For hazard identification, the team combined a simple proximity strategy with classification from a pre-trained vision transformer. Captions came from the MOLMO vision-language model with prompts tailored to rare, low-resolution hazards. The combined system cut relative error by about a third against the baselines and placed second among thirty-two teams.
Catalog record
| CREATOR | Lukas Picek, Vojtěch Čermák, Marek Hanzl |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-19944