[2002.02709] FourierNet: Compact mask representation for instance segmentation using differentiable shape decoderscontact arXivarXiv Twitter

We present FourierNet a single shot, anchor-free, fully convolutional instance segmentation method, which predicts a shape vector that is converted into contour points using a numerical transformation. Compared to previous methods, we introduce a new training technique, where we utilize a differentiable shape decoder, which achieves automatic weight balancing of the shape vector's coefficients. Fourier series was utilized as a shape encoder because of its coefficient interpretability and fast implementation. By using its lower frequencies we were able to retrieve smooth and compact masks. FourierNet shows promising results compared to polygon representation methods, achieving 30.6 mAP on the MS COCO 2017 benchmark. At lower image resolutions, it runs at 26.6 FPS with 24.3 mAP. It achieves 23.3 mAP using just 8 parameters to represent the mask, which is double the amount of parameters to predict a bounding box. Code will be available at: github.com/cogsys-tuebingen/FourierNet.

1 mentions: @yu4u
Keywords: shap
Date: 2020/02/10 12:54

Referring Tweets

@yu4u ICIP臭がする。PolarMask系single shot instance segmentation / FourierNet: Compact mask representation for instance segmentation using differentiable shape decoders / t.co/nx7krkiLKx

Related Entries

Read more 強化学習その3
0 users, 1 mentions 2019/05/07 15:47
Read more [1907.05047] BlazeFace: Sub-millisecond Neural Face Detection on Mobile GPUs
1 users, 0 mentions 2019/07/12 02:17
Read more 4th place solution | Kaggle
0 users, 1 mentions 2019/09/27 02:18
Read more DLLAB Engineer Days : ONNX Export & Optimize
0 users, 1 mentions 2019/10/14 02:18
Read more [1911.06939] Unsupervised Representation Learning for Gaze Estimationopen searchopen navigation menu...
1 users, 0 mentions 2020/06/08 09:51