diff --git a/README.md b/README.md index 3c84713..18a57ed 100644 --- a/README.md +++ b/README.md @@ -13,7 +13,7 @@ attributed to the complication in matching object queries to encoded image featu in DETR's decoder cross-attention modules.
- +
Motivated by this observation, in our paper, we propose SAM-DETR, a @@ -35,7 +35,7 @@ reference box for each object query, whose center location is used to generate corresponding position embeddings.
- +
The figure below illustrates the architecture of the appended "Semantics Aligner", which @@ -43,7 +43,7 @@ aligns the semantics of "encoded image features" and "object queries" by resampl from multiple salient points as new object queries.
- +
Being like a plug-and-play, our approach can be @@ -118,7 +118,7 @@ python test.py # unit test (should see all checking is True) Please download [COCO 2017 dataset](https://cocodataset.org/) and organize them as following: ``` -sam_detr_root/ +code_root/ ©¸©¤©¤ data/ ©¸©¤©¤ coco/ ©À©¤©¤ train2017/ @@ -134,7 +134,6 @@ sam_detr_root/ - ## Usage ### Reproducing Paper Results