Summary of Objectcompose: Evaluating Resilience Of Vision-based Models on Object-to-background Compositional Changes, by Hashmat Shadab Malik et al.
ObjectCompose: Evaluating Resilience of Vision-Based Models on Object-to-Background Compositional Changes
by Hashmat Shadab Malik, Muhammad Huzaifa, Muzammal Naseer, Salman Khan, Fahad Shahbaz Khan
First submitted to arxiv on: 7 Mar 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 method evaluates the resilience of current vision-based models against diverse object-to-background context variations. To achieve this, it harnesses the generative capabilities of text-to-image, image-to-text, and image-to-segment models to automatically generate a broad spectrum of object-to-background changes. The approach induces both natural and adversarial background changes by modifying textual prompts or optimizing latents and textual embedding of text-to-image models. It produces various versions of standard vision datasets (ImageNet, COCO), incorporating diverse and realistic backgrounds into the images or introducing color, texture, and adversarial changes in the background. Extensive experiments are conducted to analyze the robustness of vision-based models against object-to-background context variations across diverse tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about making sure that computer models used for things like recognizing objects in pictures are good at handling different situations. Imagine you’re looking at a picture of a dog, but the background is changed from a park to a city street. The model should still be able to recognize the dog correctly. The researchers came up with a way to test how well these models do in different situations by changing the background in lots of ways. They used special computer programs that can create new pictures and change the backgrounds, making it look like the dog is now standing on a mountain or something. Then they tested how well the models did at recognizing objects in these new pictures. |
Keywords
» Artificial intelligence » Embedding
