Unverified50% confidenceFactExact time
PSPNet's pyramid pooling module pools the feature map into regions of multiple different sizes (such as 1×1, 2×2, 3×3, 6×6), then upsamples the pooling results of each scale and concatenates them, to simultaneously capture local details and global contextual information.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSplatQ partitions all Gaussians into size-constrained blocks based on spatial proximity and appearance similarity.63% similarUnverifiedIn PyTorch, images are represented as four-dimensional tensors with shape [B, C, H, W] (batch, channels, height, width), and image processing operations are essentially mathematical transformations of tensors.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1028280API
curl https://kongchang.com/api/v1/knowledge/claims/1028280MCP
get_claim(id=1028280)