基於不確定性感知分布校正機制之域適應物件偵測

Domain-Adaptive Object Detection via Uncertainty-Aware Distribution Alignment

阮登科(Dang-Khoa Nguyen)

指導教授 ： 帥宏翰

國立交通大學/電機學院/電機資訊國際學位學程/碩士(2020年)

摘要

領域適應的目的是將知識從帶有註釋的源數據轉移到目標領域中幾乎沒有標籤的數據，這近年來引起了很多關注，並促進了許多多媒體應用。最近的方法顯示了使用對抗性學習通過在圖像和實例級別上對齊源圖像和目標圖像之間的分佈來減少源圖像和目標圖像之間的分佈差異的有效性。但是，這仍然具有挑戰性，因為兩個域可能具有不同的背景場景和不同的對象。此外，對象和各種圖像樣式的複雜組合會使無監督的跨域分佈對齊方式惡化。為了解決這些挑戰，在本文中，我們設計了一種用於對象檢測器無監督域自適應的端到端方法。具體來說，我們提出了一種多層次的熵注意對準（MEAA）方法，該方法包括兩個主要部分：（1）局部不確定性注意對準（LUAA）模塊，通過利用信息論來更好地感知感興趣的結構不變對象，從而增強模型通過像素域分類器的熵和（2）多級不確定性上下文對齊（MUCA）模塊來測量每個局部區域的不確定性，以基於多級域的熵來豐富相關對象的域不變信息分類器。在四個域移位對象檢測方案中評估了擬議的MEAA。實驗結果證明了在三種具有挑戰性的情況下的最新性能，以及在一個基準數據集上的競爭性能。

關鍵字

域適應 ； 物件偵測

並列摘要

Domain adaptation aims to transfer knowledge from the source data with annotations to scarcely-labeled data in the target domain, which has attracted a lot of attention in recent years and facilitated many multimedia applications. Recent approaches have shown the effectiveness of using adversarial learning to reduce the distribution discrepancy between the source and target images by aligning distribution between source and target images at both image and instance levels. However, this remains challenging since two domains may have distinct background scenes and different object. Moreover, complex combinations of objects and a variety of image styles deteriorate the unsupervised cross-domain distribution alignment. To address these challenges, in this paper, we design an end-to-end approach for unsupervised domain adaptation of object detector. Specifically, we propose a Multi-level Entropy Attention Alignment (MEAA) method that consists of two main components: (1) Local Uncertainty Attentional Alignment (LUAA) module to enhance the model better perceiving structure-invariant objects of interest by utilizing information theory to measure uncertainty of each local region via the entropy of the pixel-wise domain classifier and (2) Multi-level Uncertainty-Aware Context Alignment (MUCA) module to enrich domain-invariant information of relevant objects based on the entropy of multi-level domain classifiers. The proposed MEAA is evaluated in five domain-shift object detection scenarios. Experiment results demonstrate state-of-the-art performance on four challenging scenarios and competitive performance on one benchmark dataset.

並列關鍵字

Domain Adaptation ； Adversarial Learning ； Uncertainty-Aware ； Distribution Alignment ； Object Detection

參考文獻

[1] S. Ren, K. He, R. Girshick, and J. Sun, “Faster r-cnn: Towards real-time object detection

with region proposal networks,” in Advances in neural information processing systems,

2015, pp. 91–99.

[2] J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Imagenet: A large-scale

hierarchical image database,” in 2009 IEEE conference on computer vision and pattern