Methods › Computer Vision › Image Model Blocks › Scale Aggregation Block
Scale Aggregation Block
Introduced by Yi Li et al. in Data-Driven Neuron Allocation for Scale Aggregation Networks
archive 2025-07-28 Description, source and code snippet are the archive's method entry.
A Scale Aggregation Block concatenates feature maps at a wide range of scales. Feature maps for each scale are generated by a stack of downsampling, convolution and upsampling operations. The proposed scale aggregation block is a standard computational module which readily replaces any given transformation 𝐘=𝐓(𝐗), where 𝐗∈ℝ^(H×W×C), 𝐘∈ℝ^(H×W×Cₒ) with C and Cₒ being the input and output channel number respectively. 𝐓 is any operator such as a convolution layer or a series of convolution layers. Assume we have L scales. Each scale l is generated by sequentially conducting a downsampling 𝐃ₗ, a transformation 𝐓ₗ and an unsampling operator 𝐔ₗ:
𝐗^′ₗ=𝐃ₗ(𝐗),
𝐘^′ₗ=𝐓ₗ(𝐗^′ₗ),
𝐘ₗ=𝐔ₗ(𝐘^′ₗ),
where 𝐗^′ₗ∈ℝ^(Hₗ×Wₗ×C), 𝐘^′ₗ∈ℝ^(Hₗ×Wₗ×Cₗ), and 𝐘ₗ∈ℝ^(H×W×Cₗ). Notably, 𝐓ₗ has the similar structure as 𝐓. We can concatenate all L scales together, getting
𝐘^′=‖ᴸ₁𝐔ₗ(𝐓ₗ(𝐃ₗ(𝐗))),
where ‖ indicates concatenating feature maps along the channel dimension, and 𝐘^′ ∈ℝ^(H×W×∑ᴸ₁ Cₗ) is the final output feature maps of the scale aggregation block.
In the reference implementation, the downsampling 𝐃ₗ with factor s is implemented by a max pool layer with s×s kernel size and s stride. The upsampling 𝐔ₗ is implemented by resizing with the nearest neighbor interpolation.
Papers archive 2025-07-28
7 shown of 7, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.
-
Scale Invariance of Graph Neural Networks 28 Nov 2024 · 1 repository · arXiv:2411.19392
-
ScaleNet: Scale Invariance Learning in Directed Graphs 13 Nov 2024 · 1 repository · arXiv:2411.08758
-
ScaleNet: An Unsupervised Representation Learning Method for Limited Information 3 Oct 2023 · 0 repositories · arXiv:2310.02386
-
ScaleNet: Searching for the Model to Scale 15 Jul 2022 · 1 repository · arXiv:2207.07267Syntology ran 0 of 4 samples · 4 unverified
-
ScaleNet: A Shallow Architecture for Scale Estimation 9 Dec 2021 · 1 repository · arXiv:2112.04846
-
ScaleNAS: One-Shot Learning of Scale-Aware Representations for Visual Recognition 30 Nov 2020 · 0 repositories · arXiv:2011.14584
-
Data-Driven Neuron Allocation for Scale Aggregation Networks 20 Apr 2019 · 1 repository · arXiv:1904.09460
Tasks archive 2025-07-28
18 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.
Usage over time archive 2025-07-28
Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).
Categories archive 2025-07-28
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections