|
--- |
|
tags: |
|
- image-classification |
|
library_name: coreml |
|
license: mit |
|
--- |
|
|
|
# ResNet: Deep Residual Learning for Image Recognition |
|
|
|
ResNet introduced the concept of residual blocks and is one of the most preferred architectures for feature extraction, image classification, object detection, segmentation, and other tasks. The Core ML models in this repository correspond to the ResNet-50 variant for image classification. |
|
|
|
## Models |
|
|
|
- [Resnet50](Resnet50.mlmodel): Full precision (32 bit) model weights. |
|
- [Resnet50FP16](Resnet50FP16.mlmodel): Half precision (16 bit) model weights. |
|
- [Resnet50Int8LUT](Resnet50Int8LUT.mlmodel): Model optimized using 8-bit quantization with KMeans. |
|
- [Resnet50Headless](Resnet50Headless.mlmodel): Use as a custom image classifier base model with Create ML. |
|
|
|
## Sample Code |
|
|
|
- [Classifying Images with Vision and Core ML]( |
|
https://developer.apple.com/documentation/vision/classifying_images_with_vision_and_core_ml) |
|
|
|
## Resources |
|
|
|
- [Original source](https://github.com/fchollet/deep-learning-models/blob/master/resnet50.py) |
|
- [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) |
