Pseudo labeling (PL) is a wide-applied strategy to enlarge the labeled dataset by self-annotating the potential samples during the training process. Several works have shown that it can improve the graph learning model performance in general. However, we notice that the incorrect labels can be fatal to the graph training process. Inappropriate PL may result in the performance degrading, especially on graph data where the noise can propagate. Surprisingly, the corresponding error is seldom theoretically analyzed in the literature.

In this supplementary material, we will provide more analyses of mask prior in Section 1 and similarity transfer in Section 2. We will show the visualization results in Section 3 and the performance variance with iteration in Section 4. We will also conduct experiments to mine base categories in the target dataset in Section 5. Besides, the hyper-parameters analyses will be provided in Section 6. Finally, we will discuss the limitations in Section 7. As mentioned in Section 3.2 in the main paper, mask prior provides coarse pixel-wise category information to improve the ability of the object detection network to locate and identify objects. Our ablation studies (Table 3 in the main paper) have already proved the advantage of mask prior. To further evaluate the effectiveness of mask prior, we evaluate object detection network with/without mask generator on VOC test set. Considering that the target dataset may contain both base categories and novel categories, in which only novel categories have ground-truth bounding boxes, we evaluate our method on novel categories.

Object detection has achieved promising success, but requires large-scale fullyannotated data, which is time-consuming and labor-extensive. Therefore, we consider object detection with mixed supervision, which learns novel object categories using weak annotations with the help of full annotations of existing base object categories. Previous works using mixed supervision mainly learn the classagnostic objectness from fully-annotated categories, which can be transferred to upgrade the weak annotations to pseudo full annotations for novel categories. In this paper, we further transfer mask prior and semantic similarity to bridge the gap between novel categories and base categories. Specifically, the ability of using mask prior to help detect objects is learned from base categories and transferred to novel categories. Moreover, the semantic similarity between objects learned from base categories is transferred to denoise the pseudo full annotations for novel categories. Experimental results on three benchmark datasets demonstrate the effectiveness of our method over existing methods.

Pseudo labeling (PL) is a wide-applied strategy to enlarge the labeled dataset by self-annotating the potential samples during the training process.

Our ablation studies (Table3in the main paper) havealready proved the advantage of mask prior. From Figure 2, we can see that the coarse masks indicate the rough locations of objects which can help the object detection network predicttheboundingboxes. Tovalidate the transferability ofour similarity transfer,we evaluate our similarity network trained on COCO-60 trainval set. Wetreat the similarity prediction task as abinary classification task, in which the binary label 1 (resp., 0) means that two bounding boxes belong to the same category (resp.,different The precision, recall and F1 scores are summarized in Table 1. We observe that the gap between the performance of similarity network on base categories and novel categories is negligible (e.g., F1 Scores 84.9% v.s.