本文工作主要建立在 AAAI2018的《Meta-learning for domain generalization》,主要创新点是DICM和DSCM于crowd concept无关,并设计了特征重建和正交损失两类损失。 解决的主要问题:计数模型在未知场景的迁移能力(domain generalization) 与之前Crowd Counting中Domain Adaptive方法的不同:不依赖于未知场景的数据进行训练 本文...
进入正题,这篇文章研究的是人群计数的单领域泛化问题。人群计数(crowd counting)是个经典的计算机视觉问题,简单点说就是数出图片中含有多少人。主流方法的做法是估计出图片对应的人群密度图(density map),将图中所有像素的值相加便得到最终的计数结果,这种方法在各种数据集上都取得了不错的效果。 然而和其他视觉任务...
Domain shift across crowd data severely hinders crowd counting models to generalize to unseen scenarios. Although domain adaptive crowd counting approaches close this gap to a certain extent, they are still dependent on the target domain data to adapt (e.g. finetune) their models to the specific...
Recent deep networks have convincingly demonstrated high capability in crowd counting, which is a critical task attracting widespread attention due to its various industrial applications. Despite such progress, trained data-dependent models usually can not generalize well to unseen scenarios because of the...
PyTorch implementations of the paper: "Neuron Linear Transformation: Modeling the Domain Shift for Crowd Counting. (T-NNLS, 2021)..." domain-adaptationcrowd-counting Readme Activity 16stars 3watching 4forks Releases No releases published Packages ...
Code for the paper "Leveraging Self-Supervision for Cross-Domain Crowd Counting"(CVPR 2022) - weizheliu/Cross-Domain-Crowd-Counting
The problem of domain-adaptation in crowd counting is also addressed by training a model using the abundant labelled data available in the source domain and transferring the parameters learnt to a target domain with relatively fewer labelled data using neuron linear transformation, thereby minimizing ...
Most of the current methods treat crowd counting by density map estimation and use the Fully Convolution Network (FCN) for prediction. The mainstream framework is to predict density maps and use the sum up the density maps to get the number of people. In such methods, the main drawback is...
domain adaptationstyle transferdensity mapThe crowd counting based on the domain adaptive method is an effective unsupervised learning strategy, which does not rely on labeled samples. However, the existing methods easily cause information loss in the head region or over counting errors in the back...
Cross-Domain Adaptive Clustering for Semi-Supervised Domain Adaptation Cross-Domain Multi-Task Learning for Object Detection and Saliency Estimation Error-aware density isomorphism reconstruction for unsupervised cross-domain crowd counting Domain structure-based transfer learning for cross-domain word representat...