要解決的問題：域間差異（domain shift）導致檢測性能下降。（域可以理解成區域/不同的場景圖像）
補充：域間差異：不同的數據集具有不同的數據分佈，一般情況下訓練的模型也只能用在與這種訓練數據集分佈相似的數據集上，而用於與訓練數據集分佈不同的數據集中時，則會產生具有明顯差距的結果。也就是泛化能力。few-shot可以理解爲訓練樣本少。
方法：少鏡頭自適應方法（few-shot adaptation approach）。
優勢：一、快速適應（Fast adaptation）；二、少量數據集（Less data collection cost）；三、訓練穩定（Training stability）。
目前自適應方法面臨的問題：一、目標域數據集不足；二、目標檢測同時涉及定位和分類，進一步複雜化了模型自適應過程；三、存在過適應的問題。
圖像級差異：整個場景圖像的差異；實例級差異：目標間的差異。
做法：we first introduce a pairing mechanism over source and target features to alleviate the issue of insufficient target domain samples. We then propose a bi-level module to adapt the source trained detector to the target domain: 1) the split pooling based image level adaptation module uniformly extracts and aligns paired local patch features over locations, with different scale and aspect ratio; 2) the instance level adaptation module semantically aligns paired object features while avoids inter-class confusion. Meanwhile, a source model feature regularization (SMFR) is applied to stabilize the adaptation process of the two modules.Combining these contributions gives a novel few-shot adaptive Faster-RCNN framework, termed FAFRCNN, which effectively adapts to target domain with a few labeled samples. Experiments with multiple datasets show that our model achieves new state-of-the-art performance under both the interested few-shot domain adaptation(FDA) and unsupervised domain adaptation(UDA) setting.
我們首先在源（暫時理解爲源圖像）和目標特徵上引入配對機制，以緩解目標域樣本不足的問題。然後，我們提出了一個雙層模塊，使源訓練檢測器適應目標域：1）基於分割池的圖像級自適應模塊在不同的尺度和縱橫比下，在不同的位置上均勻地提取和對齊成對的局部塊特徵；2）實例級自適應模塊在語義上對齊成對的對象特徵，同時避免類間混淆。同時，應用源模型特徵正則化（SMFR）來穩定這兩個模塊的自適應過程，結合這些貢獻，提出了一種新的多鏡頭自適應Faster-RCNN框架，稱爲FAFRCNN，該框架通過少量標記樣本有效地適應目標域。對多個數據集的實驗表明，該模型在感興趣的少鏡頭域自適應（FDA）和無監督域自適應（UDA）兩種情況下都取得了最新的性能。
Datasets：The SIM10K dataset contains 10k synthetic images with bounding box annotation for car, motorbike and person. The Cityscapes dataset contains around 5000 accurately annotated real world images with pixel-level category labels. The Foggy Cityscapes dataset is generated from Cityscapes with simulated fog. The Udacity self-driving dataset (Udacity for short) is an open source dataset collected with different illumination, camera condition and surroundings as Cityscapes.