</> Code with Papers by Syntology

Methods › Computer Vision › Image Model Blocks › Split Attention

Split Attention

20 papers tagged archive 2025-07-28

archive 2025-07-28 Description, source and code snippet are the archive's method entry.

A Split Attention block enables attention across feature-map groups. As in ResNeXt blocks, the feature can be divided into several groups, and the number of feature-map groups is given by a cardinality hyperparameter K. The resulting feature-map groups are called cardinal groups. Split Attention blocks introduce a new radix hyperparameter R that indicates the number of splits within a cardinal group, so the total number of feature groups is G = KR. We may apply a series of transformations {ℱ₁, ℱ₂, ⋯ℱ_G} to each individual group, then the intermediate representation of each group is Uᵢ = ℱᵢ(X), for i ∈ {1, 2, ⋯G}.

A combined representation for each cardinal group can be obtained by fusing via an element-wise summation across multiple splits. The representation for k-th cardinal group is Ûᵏ = ∑_(j=R(k-1)+1)^(R k) Uⱼ, where Ûᵏ ∈ℝ^(H×W×C/K) for k∈1,2,...K, and H, W and C are the block output feature-map sizes. Global contextual information with embedded channel-wise statistics can be gathered with global average pooling across spatial dimensions sᵏ∈ℝ^(C/K). Here the c-th component is calculated as:

sᵏ_c = 1/(H×W) ∑ᵢ₌₁ᴴ∑ⱼ₌₁ᵂ Ûᵏ_c(i, j).

A weighted fusion of the cardinal group representation Vᵏ∈ℝ^(H×W×C/K) is aggregated using channel-wise soft attention, where each feature-map channel is produced using a weighted combination over splits. The c-th channel is calculated as:

Vᵏ_c=∑ᵢ₌₁ᴿ aᵏᵢ(c) U_(R(k-1)+i) ,

where aᵢᵏ(c) denotes a (soft) assignment weight given by:

aᵢᵏ(c) = (exp(𝒢ᶜᵢ(sᵏ)))/(∑ⱼ₌₀ᴿ exp(𝒢ᶜⱼ(sᵏ))) if R>1, 1/(1+exp(-𝒢ᶜᵢ(sᵏ))) if R=1,

and mapping 𝒢ᵢᶜ determines the weight of each split for the c-th channel based on the global context representation sᵏ.

Source: ResNeSt: Split-Attention NetworksSee Code · zhanghang1989/ResNeSt

Papers archive 2025-07-28

20 shown of 20, 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.

Tasks archive 2025-07-28

20 shown of 45 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.

TaskPapers
Semantic Segmentation5
Instance Segmentation4
Segmentation4
Image Classification3
Object Detection3
Transfer Learning3
image-classification3
object-detection2
Action Recognition1
Anomaly Detection1
Articles1
Computational Efficiency1
Decoder1
Domain Adaptation1
ECG Classification1
Earth Observation1
Emotion Recognition1
Experimental Design1
Face Detection1
Face Hallucination1

Usage over time archive 2025-07-28

Papers per year tagged with Split Attention: 2020 to 2024, peak 6 6 0 2020: 5 papers 2020 2021: 6 papers 2021 2022: 2 papers 2022 2023: 4 papers 2023 2024: 3 papers 2024
Papers per year the archive tags with this method, by the paper's archive date (20 dated). Bars are counts, not a trend claim.

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

Image Model Blocks

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