☆ 3.8 Proceedings Paper

Bi-level Alignment for Cross-Domain Crowd Counting

2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR) (2022)

Pixel-Wise Crowd Understanding via Synthetic Data

Qi Wang et al.

Summary: This paper explores crowd analysis using computer vision techniques, focusing on pixel-wise crowd understanding. By developing a synthetic dataset called GCC Dataset and proposing two methods to improve crowd understanding, the study aims to achieve better performance in real-world scenarios.

INTERNATIONAL JOURNAL OF COMPUTER VISION (2021)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

LightTrack: Finding Lightweight Neural Networks for Object Tracking via One-Shot Architecture Search

Bin Yan et al.

Summary: Object tracking has made significant progress in recent years, but state-of-the-art trackers are becoming heavier and more expensive. LightTrack, utilizing neural architecture search, has been designed to be more lightweight and efficient, with experimental results showing its effectiveness in resource-constrained environments.

2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 (2021)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

One-Shot Neural Ensemble Architecture Search by Diversity-Guided Search Space Shrinking

Minghao Chen et al.

Summary: The study introduces a neural ensemble architecture search method that simultaneously searches for multiple diverse models to address the limitations in accuracy and robustness of single models. By introducing a novel diversity-based metric and enabling a new search dimension, the challenges of enlarged search space and increased complexity for the searched model are successfully tackled.

2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 (2021)

添加到收藏夹

Proceedings Paper Acoustics