oxicuda-vision 0.2.0

Vision Transformer & CLIP primitives for OxiCUDA: ViT patch embedding, multi-head self-attention, CLIP contrastive learning, FPN, RoI align, DETR decoder — pure Rust, zero CUDA SDK dependency.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12

//! Feature Pyramid Network (FPN) components.
//!
//! Provides:
//! - **`LateralConv1x1`**: 1×1 lateral convolution for channel reduction.
//! - **`Fpn`**: top-down feature pyramid with nearest-neighbour upsampling
//!   and 3×3 smoothing convolutions.

pub mod lateral;
pub mod top_down;

pub use lateral::{LateralConv1x1, LateralWeights};
pub use top_down::{FeatureMap, Fpn, FpnConfig};