shenyunhang/APE
A foundational vision-language model for universal visual perception that detects and segments objects using natural language descriptions across diverse domains.
Not currently ranked — collecting fresh signals.
star history
APE aligns visual and language representations through an innovative prompting mechanism to enable universal visual perception. The model performs open-world object detection, instance segmentation, semantic segmentation, and referring expression comprehension using a single unified architecture. It achieves competitive performance across 160 datasets by leveraging vision-language transformer foundations with thousands of vocabulary entries and language descriptions for flexible perception in varied scenarios.
Frequently asked
- What is shenyunhang/APE?
- A foundational vision-language model for universal visual perception that detects and segments objects using natural language descriptions across diverse domains.
- Is APE open source?
- Yes — shenyunhang/APE is open source, released under the Apache-2.0 license.
- What language is APE written in?
- shenyunhang/APE is primarily written in Python.
- How popular is APE?
- shenyunhang/APE has 609 stars on GitHub.
- Where can I find APE?
- shenyunhang/APE is on GitHub at https://github.com/shenyunhang/APE.