Weakly Supervised Semantic Segmentation by Pixel-to-Prototype Contrast [PDF]
Though image-level weakly supervised semantic seg-mentation (WSSS) has achieved great progress with Class Activation Maps (CAMs) as the cornerstone, the large su-pervision gap between classification and segmentation still hampers the model to generate ...
Ye Du +3 more
semanticscholar +1 more source
Intermediate Prototype Mining Transformer for Few-Shot Semantic Segmentation [PDF]
Few-shot semantic segmentation aims to segment the target objects in query under the condition of a few annotated support images. Most previous works strive to mine more effective category information from the support to match with the corresponding ...
Yuanwei Liu +3 more
semanticscholar +1 more source
Dynamic Prototype Mask for Occluded Person Re-Identification [PDF]
Although person re-identification has achieved an impressive improvement in recent years, the common occlusion case caused by different obstacles is still an unsettled issue in real application scenarios.
Lei Tan +3 more
semanticscholar +1 more source
Neural Prototype Trees for Interpretable Fine-grained Image Recognition [PDF]
Prototype-based methods use interpretable representations to address the black-box nature of deep learning models, in contrast to post-hoc explanation methods that only approximate such models.
Meike Nauta, Ron van Bree, C. Seifert
semanticscholar +1 more source
SARS-CoV-2 mRNA Vaccine Design Enabled by Prototype Pathogen Preparedness
A vaccine for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is needed to control the coronavirus disease 2019 (COVID-19) global pandemic.
Kizzmekia S. Corbett +61 more
semanticscholar +1 more source
Cross-Domain Detection via Graph-Induced Prototype Alignment [PDF]
Applying the knowledge of an object detector trained on a specific domain directly onto a new domain is risky, as the gap between two domains can severely degrade model's performance.
Minghao Xu +4 more
semanticscholar +1 more source
Universal-Prototype Enhancing for Few-Shot Object Detection [PDF]
Few-shot object detection (FSOD) aims to strengthen the performance of novel object detection with few labeled samples. To alleviate the constraint of few samples, enhancing the generalization ability of learned features for novel objects plays a key ...
Aming Wu +3 more
semanticscholar +1 more source
Design, fabrication and test of integrated micro-scale vibration based electromagnetic generator [PDF]
This paper discusses the design, fabrication and testing of electromagnetic microgenerators. Three different designs of power generators are partially microfabricated and assembled.
Beeby +18 more
core +1 more source
Bayah Dome Geopark Information System Based On Website Using Prototype Method
To realize a national Geopark and a World Geopark, it is necessary to socialize and disseminate information that can be accessed by the people of Lebak Regency in particular and the wider community in general to introduce all the geological heritage that
Dentik Karyaningsih +3 more
doaj +1 more source
Continual Prototype Evolution: Learning Online from Non-Stationary Data Streams [PDF]
Attaining prototypical features to represent class distributions is well established in representation learning. However, learning prototypes online from streaming data proves a challenging endeavor as they rapidly become outdated, caused by an ever ...
Matthias De Lange, T. Tuytelaars
semanticscholar +1 more source

