object detection
What is it about?
In order to strengthen feature fusion and solve the two problems of feature redundancy and high abstraction, Multi-scale Redistribution Feature Pyramid Network (MRFPN) designs modified-BiFPN. The features output by the modified-BiFPN module are semantically balanced through the balanced feature map, so as to alleviate the semantic differences between multi-scales.Then a new channel attention module is proposed, which realizes the multi-scale association of the feature information fused to the balanced feature map. Finally, a new feature pyramid is formed through the residual edge for prediction
Why is it important?
We have designed some new modules in MRFPN for object detection. The main contributions are as follows: (1) modified-BiFPN solves the problem of single form of multi-scale feature fusion and alleviates the problem of feature redundancy; (2) The channel assignment mechanism(CAM) restores the feature pyramid model, which provides a new direction for feature prediction.