InceptionNeXt: When Inception Meets ConvNeXt
By Weihao Yu, Pan Zhou, Shuicheng Yan, Xinchao Wang. Computer Vision and Pattern Recognition, 2023
350 citations recorded by the source corpus.
Inspired by the long-range modeling ability of ViTs, large-kernel convolutions are widely studied and adopted recently to enlarge the receptive field and improve model performance, like the remarkable work ConvNeXt which employs 7 × 7 depthwise convolution. Although such depth-wise operator only consumes a few FLOPs, it largely harms the model efficiency on powerful computing devices due to the high memory access costs. For example, ConvNeXt-T has similar FLOPs with ResNet-50 but only achieves ~60% throughputs when trained on A100 GPUs with full precision. Although reducing the kernel size of ConvNeXt can improve speed, it results in significant performance degradation, which poses a challenging problem: How to speed up large-kernel-based CNN models while preserving their performance. To tackle this issue, inspired by Inceptions, we propose to decompose large-kernel depth-wise convolution into four parallel branches along channel dimension, i.e., small square kernel, two orthogonal band kernels, and an identity mapping. With this new Inception depthwise convolution, we build a series of networks, namely IncepitonNeXt, which not only enjoy high throughputs but also maintain competitive performance. For instance, InceptionNeXt-T achieves 1.6 × higher training throughputs than ConvNeX-T, as well as attains 0.2% top-1 accuracy improvement on ImageNet-1K. We antici-pate InceptionNeXt can serve as an economical baseline for future architecture design to reduce carbon footprint.
Subfields: Computational Theory and Mathematics, Information Systems, Information Systems and Management
View Paper: https://www.semanticscholar.org/paper/182e6c877155d7730378846380a3d1dc294fb54a
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
Citations in the corpus
- Poly Kernel Inception Network for Remote Sensing Detection (2024, background): “2(d), PKI Module is an inception-style module [57, 77] that comprises a small-kernel convolution to grasp local information, followed by a set of parallel depth-wise convolutions to capture contextual information across multiple scales.”
- MambaOut: Do We Really Need Mamba for Vision?* (2024, neutral): “The meta-architecture of Gated CNN and Mamba can be considered as a simplified integration of the MetaFormer’s [99] token mixer and an MLP, akin to MetaNeXt [101].”
- SCSA: Exploring the Synergistic Effects Between Spatial and Channel Attention (2024, neutral): “Additionally, to fully interact with the different sub-features decomposed in SMSA, We select to implement a simpler single-head self-attention mechanism instead of the combination of multi-head self-attention with channel shuffling [66].”