Dense Prediction

التنبؤ الكثيف

مهام تتطلب إنتاج تنبؤ لكل بكسل في الصورة (كالتجزئة الدلالية وتقدير العمق) بدلاً من تصنيف واحد للصورة كلها.

Tasks that require producing a prediction for every pixel in the image (such as semantic segmentation and depth estimation) rather than a single classification for the whole image.

Also translated asالتنبؤ على مستوى البكسل، التنبؤ الشامل

First appears in this corpus in: Fully Convolutional Networks for Semantic Segmentation (2015)

Appears in these papers

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs2017 in the sky ✦
DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs2017 in the sky ✦
Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data2024 in the sky ✦
Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data2024 in the sky ✦
Fully Convolutional Networks for Semantic Segmentation2015 in the sky ✦
Fully Convolutional Networks for Semantic Segmentation2015 in the sky ✦
RAFT: Recurrent All-Pairs Field Transforms for Optical Flow2020 in the sky ✦
SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers2021 in the sky ✦
Swin Transformer: Hierarchical Vision Transformer Using Shifted Windows2021 in the sky ✦
Swin Transformer: Hierarchical Vision Transformer Using Shifted Windows2021 in the sky ✦