Statements (61)
| Predicate | Object |
|---|---|
| gptkbp:instance_of |
gptkb:neural_networks
|
| gptkbp:allows |
very deep networks
|
| gptkbp:developed_by |
gptkb:Kaiming_He
|
| gptkbp:features |
residual connections
|
| gptkbp:has_applications_in |
gptkb:medical_imaging
object detection video analysis face recognition |
| gptkbp:has_variants |
gptkb:Res_Net-101
gptkb:Res_Net-152 gptkb:Res_Net-50 |
| https://www.w3.org/2000/01/rdf-schema#label |
Res Net architecture
|
| gptkbp:improves |
training of deep networks
|
| gptkbp:influenced |
subsequent architectures
|
| gptkbp:introduced_in |
gptkb:2015
|
| gptkbp:is_based_on |
fully connected layers
batch normalization convolutional layers skip connections Re LU activation function |
| gptkbp:is_characterized_by |
identity mapping
|
| gptkbp:is_evaluated_by |
gptkb:CIFAR-10
gptkb:Pascal_VOC gptkb:Celeb_A_dataset gptkb:COCO_dataset gptkb:Image_Net_dataset gptkb:CIFAR-100 LFW dataset Oxford Pets dataset SVHN dataset |
| gptkbp:is_implemented_in |
gptkb:Tensor_Flow
gptkb:Py_Torch |
| gptkbp:is_known_for |
high accuracy
modular design scalability improving generalization reducing overfitting efficient training flexibility in architecture design solving vanishing gradient problem |
| gptkbp:is_part_of |
gptkb:AI_technology
image processing techniques deep learning frameworks machine learning frameworks computer vision research |
| gptkbp:is_popular_in |
deep learning community
|
| gptkbp:is_used_in |
gptkb:vehicles
gptkb:robotics augmented reality real-time applications transfer learning feature extraction image segmentation generative models style transfer computer vision tasks smart surveillance |
| gptkbp:used_for |
image classification
|
| gptkbp:won |
gptkb:ILSVRC_2015
|
| gptkbp:bfsParent |
gptkb:Deep_Lab
|
| gptkbp:bfsLayer |
5
|