PASCAL-5i

The PASCAL-5i dataset is a benchmark that contains a subset of the PASCAL VOC dataset, specifically designed for evaluating open-vocabulary semantic segmentation by providing a few annotated images for each of the novel categories.

Papers using PASCAL-5i (6)

Probabilistic Prototype Calibration of Vision-Language Models for Generalized Few-shot Semantic Segmentation2025Open-vocabulary Semantic Segmentation with Frozen Vision-Language Models2022 · 10 citesFew-Shot Image Classification and Segmentation as Visual Question Answering Using Vision-Language Models2024 · 2 citesUniBoost: Unsupervised Unimodal Pre-training for Boosting Zero-shot Vision-Language Tasks2023 · 1 citesMulti-Modal Prototypes for Open-World Semantic Segmentation2023AFANet: Adaptive Frequency-Aware Network for Weakly-Supervised Few-Shot Semantic Segmentation2024