Ảnh bìa

Dynamic Curriculum Learning for Great Ape Detection in the Wild

CC BY

Lưu vào:
Hiển thị chi tiết
Tác giả chính: Xinyu, Yang, Tilo, Burghardt, Majid, Mirmehdi
Định dạng: Sách
Ngôn ngữ:English
Nhà xuất bản: Springer 2023
Chủ đề:
We propose a novel end-to-end curriculum learning approach for sparsely labelled animal datasets leveraging large volumes of unlabelled data to improve supervised species detectors. We exemplify the method in detail on the task of finding great apes in camera trap footage taken in challenging real-world jungle environments. In contrast to previous semi-supervised methods, our approach adjusts learning parameters dynamically over time and gradually improves detection quality by steering training towards virtuous self-reinforcement. To achieve this, we propose integrating pseudo-labelling with curriculum learning policies and show how learning collapse can be avoided. We discuss theoretical arguments, ablations, and significant performance improvements against various state-of-the-art systems when evaluating on the Extended PanAfrican Dataset holding approx. 1.8M frames.
Truy cập trực tuyến:https://link.springer.com/article/10.1007/s11263-023-01748-3
https://dlib.phenikaa-uni.edu.vn/handle/PNK/8268
Từ khóa: Thêm từ khóa
Không có từ khóa, Hãy là người đầu tiên đánh dấu biểu ghi này!

Tài liệu tương tự

Tài liệu tương tự