Understanding the Spatial Attention Module (ThunderNet)
The Spatial Attention Module, also known as SAM, is a critical component of ThunderNet, an object detection feature extraction module. The SAM is designed to adjust the feature distribution of the feature map accurately by making use of knowledge from RPN. In this article, we will go over the math behind the SAM, its structure, and its functions in detail.
The Concept behind SAM
The SAM is a feature extraction module that shares knowled
Overview of Spatio-Temporal Features Extraction
If you're interested in understanding how things move, then you've likely come across the term "spatio-temporal" before. This refers to anything that has both a spatial (where) and a temporal (when) component to it. By analyzing these components, we can extract features that tell us a lot about how things move and change over time.
One important use of spatio-temporal features extraction is in the field of stability measurement. Essentially, this
What is TUM?
TUM stands for Thinned U-Shape Module, which is a feature extraction block used for object detection models. It is a newer structure that was introduced as part of M2Det architecture.
How is TUM Different from Other Feature Extraction Blocks?
TUM differs from other feature extraction blocks, such as FPN and RetinaNet, by adopting a thinner U-shape structure. The encoder is a series of 3x3 convolution layers with stride 2, while the decoder takes the outputs of these layers as it
Overview of TridentNet Block:
The TridentNet Block is a feature extractor that is utilized in object detection models. Through this block, the backbone network adapts to different scales to generate multiple scale-specific feature maps. This is achieved by utilizing dilated convolutions, where the different branches of the trident block share the same network structure and parameters, but have different receptive fields.
Understanding TridentNet Block:
Object detection models are a type of c