What is the method?
- Use high-resolution images to capture affluent spatial information .
- Use downsampling to capture large receptive field.
3/27/25
What is the method?
- Let low-resolution images go through the full semantic perception network.
- Integrate medium and high resolution features with limited parameters.
3/27/25
What is the method?
- Build multi-scale CNNs on Laplacian pyramid.
- Use temporal consistent superpixels to avoid flickering.
3/27/25
What is the idea?
Motivated by LLM's strong zero-shot generalisation capabilities. In this work, our goal is to build a foundation model for
image segmentation. That is, we seek to develop a prompt-
able model and pre-train it on a broad dataset using a task
that enables powerful generalisation.
3/27/25