EdaDet: Open-Vocabulary Object Detection Using Early Dense Alignment
About
Vision-language models such as CLIP have boosted the performance of open-vocabulary object detection, where the detector is trained on base categories but required to detect novel categories. Existing methods leverage CLIP's strong zero-shot recognition ability to align object-level embeddings with textual embeddings of categories. However, we observe that using CLIP for object-level alignment results in overfitting to base categories, i.e., novel categories most similar to base categories have particularly poor performance as they are recognized as similar base categories. In this paper, we first identify that the loss of critical fine-grained local image semantics hinders existing methods from attaining strong base-to-novel generalization. Then, we propose Early Dense Alignment (EDA) to bridge the gap between generalizable local semantics and object-level prediction. In EDA, we use object-level supervision to learn the dense-level rather than object-level alignment to maintain the local fine-grained semantics. Extensive experiments demonstrate our superior performance to competing approaches under the same strict setting and without using external training resources, i.e., improving the +8.4% novel box AP50 on COCO and +3.9% rare mask AP on LVIS.