WideResNet101: Image Classification
WideResNet101 is a high-performance variant of residual networks, boosting model capacity by significantly increasing network width (channel count) rather than adding layers. Building on ResNet-101, it employs wider residual blocks (e.g., width factors of 2 or 4) to expand feature dimensions for enhanced local detail capture, while maintaining shallower depth to mitigate gradient vanishing. Inheriting residual skip connections and batch normalization, it ensures stable training and fast convergence, achieving higher accuracy than ResNet-101 on datasets like ImageNet. Despite moderate parameter growth, optimized computational efficiency makes it suitable for high-precision tasks (e.g., image classification, object detection), balancing performance and resource constraints.
Source model
- Input shape: 224x224
- Number of parameters: 121.01M
- Model size: 483.82M
- Output shape: 1x1000
The source model can be found here
Performance Reference
Please search model by model name in Model Farm
Inference & Model Conversion
Please search model by model name in Model Farm
License
Source Model: BSD-3-CLAUSE
Deployable Model: APLUX-MODEL-FARM-LICENSE