The Squeeze-and-Excitation Block is an architectural unit designed to improve the representational power of a network by enabling it to perform dynamic channel-wise feature recalibration. The process is:
Paper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Image Classification | 104 | 14.09% |
Object Detection | 48 | 6.50% |
Semantic Segmentation | 38 | 5.15% |
Classification | 37 | 5.01% |
General Classification | 29 | 3.93% |
Instance Segmentation | 17 | 2.30% |
Quantization | 13 | 1.76% |
Multi-Task Learning | 10 | 1.36% |
Image Segmentation | 7 | 0.95% |
Component | Type |
|
---|---|---|
Average Pooling
|
Pooling Operations | |
Convolution
|
Convolutions | |
Dense Connections
|
Feedforward Networks | |
ReLU
|
Activation Functions | |
Sigmoid Activation
|
Activation Functions |