Adapter-style efficient transfer learning (ETL) has shown excellent performance in the tuning
of vision-language models (VLMs) under the low-data regime, where only a few additional …

Large-scale vision-language models (VLMs) pre-trained on billion-level data have learned
general visual representations and broad visual concepts. In principle, the well-learned …

Modern visual recognition models often display overconfidence due to their reliance on
complex deep neural networks and one-hot target supervision, resulting in unreliable …

Semantic segmentation models comprise an encoder to extract features and a classifier for
prediction. However, the learning of the classifier suffers from the ambiguity which is caused …

Few-shot object detection (FSOD) identifies objects from extremely few annotated samples.
Most existing FSOD methods, recently, apply the two-stage learning paradigm, which …

Fine-tuning pre-trained vision-language models (VLMs), eg, CLIP, for the open-world
generalization has gained increasing popularity due to its practical value. However …

Retrieving natural images with the query sketches under the zero-shot scenario is known as
zero-shot sketch-based image retrieval (ZS-SBIR). Most of the best-performing methods …

We address a weakly-supervised low-shot instance segmentation, an annotation-efficient
training method to deal with novel classes effectively. Since it is an under-explored problem …

Few-shot segmentation remains challenging due to the limitations of its labeling information
for unseen classes. Most previous approaches rely on extracting high-level feature maps …

Land-cover mapping is one of the vital applications in Earth observation aiming at
classifying each pixel's land-cover type of remote-sensing images. As natural and human …