Rfdetr Medium Fp16 Coreml
RF-DETR Medium model exported to CoreML format (FP16) for Apple devices. Best accuracy with 576×576 resolution.
Model Details
- Model Type: Object Detection
- Framework: CoreML
- Precision: FP16
- Input Resolution: 576×576
- Performance: ~24 FPS on iPhone 13 Mini
- Inference Time: 41.85 ms
Usage
This model can be used for real-time object detection on Apple devices (iOS, macOS, watchOS, tvOS).
Custom conversion to CoreML
Check the repo https://github.com/pzoltowski/rf-detr-coreml
Performance
| Model | Resolution | Inference Time (ms) | FPS |
|---|---|---|---|
| RF-DETR-Nano | 384×384 | 17.96 | ~56 |
| RF-DETR-Small | 512×512 | 27.07 | ~37 |
| RF-DETR-Medium | 576×576 | 41.85 | ~24 |
Performance measured on iPhone 13 Mini using Xcode performance tools with FP16 precision.
- Downloads last month
- 5