Book
Involution: Inverting the Inherence of Convolution for Visual Recognition
| Identifier | --duolijiehu-1740 |
|---|---|
| Penulis / Pengarang | Duo Li, Jie Hu, Changhu Wang |
| Kata Kunci | inverting the inherence duo li |
| Tanggal Publikasi | 10 Mar 2021 |
| Bahasa | Bahasa Inggris |
| Abstrak / Deskripsi | Convolution has been the core ingredient of modern neu- ral networks, triggering the surge of deep learning in vi- sion. In this work, we rethink the inherent principles of standard convolution for vision tasks, speciï¬Âcally spatial- agnostic and channel-speciï¬Âc. Instead, we present a novel atomic operation for deep neural networks by inverting the aforementioned design principles of convolution, coined as involution. We additionally demystify the recent pop- ular self-attention operator and subsume it into our invo- lution family as an over-complicated instantiation. The proposed involution operator could be leveraged as fun- damental bricks to build the new generation of neural net- works for visual recognition, powering different deep learn- ing models on several prevalent benchmarks, including Im- ageNet classiï¬Âcation, COCO detection and segmentation, together with Cityscapes segmentation. Our involution- based models improve the performance of convolutional baselines using ResNet-50 by up to 1.6% top-1 accuracy, 2.5% and 2.4% bounding box AP, and 4.7% mean IoU ab- solutely while compressing the computational cost to 66%, 65%, 72%, and 57% on the above benchmarks, respectively. Code and pre-trained models for all the tasks are available at https://github.com/d-li14/involution. |
Attached Files / Download (1)
| No | Nama Berkas | Format | Ukuran | Aksi |
|---|---|---|---|---|
| 1 | 2103.06255.pdf | application/pdf | 4,6 MB | ⬇️ Buka / Download |
Comments & Discussion (0)
Belum ada komentar untuk koleksi ini. Jadilah yang pertama berdiskusi!