Abstract:Machine learning algorithms are commonly used for quickly and efficiently counting people from a crowd. Test-time adaptation methods for crowd counting adjust model parameters and employ additional data augmentation to better adapt the model to the specific conditions encountered during testing. The majority of current studies concentrate on unsupervised domain adaptation. These approaches commonly perform hundreds of epochs of training iterations, requiring a sizable number of unannotated data of every new target domain apart from annotated data of the source domain. Unlike these methods, we propose a meta-test-time adaptive crowd counting approach called CrowdTTA, which integrates the concept of test-time adaptation into the meta-learning framework and makes it easier for the counting model to adapt to the unknown test distributions. To facilitate the reliable supervision signal at the pixel level, we introduce uncertainty by inserting the dropout layer into the counting model. The uncertainty is then used to generate valuable pseudo labels, serving as effective supervisory signals for adapting the model. In the context of meta-learning, one image can be regarded as one task for crowd counting. In each iteration, our approach is a dual-level optimization process. In the inner update, we employ a self-supervised consistency loss function to optimize the model so as to simulate the parameters update process that occurs during the test phase. In the outer update, we authentically update the parameters based on the image with ground truth, improving the model's performance and making the pseudo labels more accurate in the next iteration. At test time, the input image is used for adapting the model before testing the image. In comparison to various supervised learning and domain adaptation methods, our results via extensive experiments on diverse datasets showcase the general adaptive capability of our approach across datasets with varying crowd densities and scales.

Few-Shot Scene Adaptive Crowd Counting Using Meta-Learning

AdaCrowd: Unlabeled Scene Adaptation for Crowd Counting

Meta-Knowledge and Multi-Task Learning-Based Multi-Scene Adaptive Crowd Counting

Unlabeled scene adaptive crowd counting via meta-ensemble learning

Crowd Counting Using Meta-Test-Time Adaptation

MetaUSACC: Unlabeled scene adaptation for crowd counting via meta-auxiliary learning

Learn from Object Counting: Crowd Counting with Meta-learning

Scene-adaptive crowd counting method based on meta learning with dual-input network DMNet

Few-Shot Crowd Counting Via Self-supervised Learning

Cross Domain Adaptation of Crowd Counting with Model-Agnostic Meta-Learning

Few-Shot Classification of Aerial Scene Images via Meta-Learning

LEVERAGE MULTI-SCALE DILATED CONVOLUTIONAL NEURAL NETWORK WITH GLOBAL ATTENTION FEATURE FUSION FOR CROWD COUNTING

Crowd Counting with Density Adaption Networks

A Hybrid Approach with Optimization-Based and Metric-Based Meta-Learner for Few-Shot Learning

Weakly-Supervised Scene-Specific Crowd Counting Using Real-Synthetic Hybrid Data

Few-shot Learning with Meta Metric Learners

Adaptive Context Learning Network for Crowd Counting.

Fine-grained Domain Adaptive Crowd Counting Via Point-derived Segmentation

Scene-Adaptive Attention Network for Crowd Counting

Crowd Counting With Limited Labeling Through Submodular Frame Selection

A metric-based meta-learning approach combined attention mechanism and ensemble learning for few-shot learning