SAM-Aware Graph Prompt Reasoning Network for Cross-Domain Few-Shot Segmentation

Authors: Shi-Feng Peng, Guolei Sun, Yong Li, Hongsong Wang, Guo-Sen Xie

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on four standard CD-FSS datasets demonstrate that our method establishes new state-of-the-art results. [...] Experiments, Datasets and Implementation Details, Comparisons with State-of-the-arts, Ablation Study and Visualizations
Researcher Affiliation Academia 1School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China 2Computer Vision Laboratory, ETH Zurich 3School of Computer Science and Engineering, Southeast University, Nanjing, China EMAIL; EMAIL; EMAIL; EMAIL
Pseudocode No The paper describes the method conceptually and mathematically through equations but does not present a structured pseudocode or algorithm block.
Open Source Code Yes Code https://github.com/CVL-hub/GPRN.
Open Datasets Yes We fine-tune and test it on four datasets: Deepglobe (Demir et al. 2018), ISIC (Codella et al. 2019), Chest X-Ray (Candemir et al. 2013), and FSS-1000 (Li et al. 2020).
Dataset Splits Yes During fine-tuning, we used 120, 520, 60, and 20 samples from the Deepglobe, FSS-1000, ISIC, and Chest X-ray datasets, respectively. For evaluation, we randomly sample 600 episodes from ISIC, Deepglobe, and Chest X-Ray, and 2400 episodes from FSS-1000, averaging the results over five random seeds.
Hardware Specification No The paper mentions using a ResNet-50 backbone but does not provide specific details about the hardware (GPU/CPU models, memory, etc.) used for conducting experiments.
Software Dependencies No The paper mentions using SSP (Fan et al. 2022) and a ResNet-50 backbone, but does not provide specific software dependencies with version numbers (e.g., Python, PyTorch, CUDA versions).
Experiment Setup Yes All images are initially resized to 400x400 for input into the CNN and subsequently upsampled to 1024x1024 to meet SAM’s input requirements. The number of masks l is fixed as 40. ... The parameter α is set to 0.1. The numbers of positive and negative point prompts, |Z| and |N|, are both set to 20, and β is set to 0.5. During training, our model is optimized with 0.9 momentum and an initial learning rate of 1e-3 and training with 5 epochs. ... We employed the SGD optimizer with a learning rate of 5e-4, momentum of 0.9, and weight decay of 5e-4.