Update README.md
Browse files
README.md
CHANGED
|
@@ -35,6 +35,23 @@ The model's weights can be found [here](./yolo11m_urchin_trained.pt) | Also avai
|
|
| 35 |
- **[TorchScript (best.torchscript)](./train/weights/best.torchscript)**
|
| 36 |
- **[NCNN](./train/weights/best_ncnn_model/model.ncnn.bin)**: Efficient for mobile platforms and embedded systems.
|
| 37 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
# Intended Use
|
| 39 |
- Real-time detections on underwater footage
|
| 40 |
- Post-processed video/imagery for detecting sea urchins in underwater environments
|
|
|
|
| 35 |
- **[TorchScript (best.torchscript)](./train/weights/best.torchscript)**
|
| 36 |
- **[NCNN](./train/weights/best_ncnn_model/model.ncnn.bin)**: Efficient for mobile platforms and embedded systems.
|
| 37 |
|
| 38 |
+
## Benchmarks
|
| 39 |
+
|index|Format|Status❔|Size \(MB\)|metrics/mAP50-95\(B\)|Inference time \(ms/im\)|FPS|
|
| 40 |
+
|---|---|---|---|---|---|---|
|
| 41 |
+
|0|PyTorch|✅|38\.7|0\.5694|23\.69|42\.21|
|
| 42 |
+
|1|TorchScript|✅|77\.1|0\.5619|21\.34|46\.86|
|
| 43 |
+
|2|ONNX|❎|76\.7|NaN|NaN|NaN|
|
| 44 |
+
|3|OpenVINO|❌|0\.0|NaN|NaN|NaN|
|
| 45 |
+
|4|TensorRT|✅|86\.4|0\.5619|20\.72|48\.27|
|
| 46 |
+
|5|CoreML|❌|0\.0|NaN|NaN|NaN|
|
| 47 |
+
|6|TensorFlow SavedModel|✅|193\.1|0\.5619|40\.84|24\.48|
|
| 48 |
+
|7|TensorFlow GraphDef|✅|76\.8|0\.5619|38\.65|25\.88|
|
| 49 |
+
|11|PaddlePaddle|✅|153\.4|0\.5619|30\.63|32\.65|
|
| 50 |
+
|12|MNN|✅|76\.6|0\.5615|1133\.73|0\.88|
|
| 51 |
+
|13|NCNN|✅|76\.6|0\.5619|1068\.97|0\.94|
|
| 52 |
+
|14|IMX|❌|0\.0|NaN|NaN|NaN|
|
| 53 |
+
|15|RKNN|❌|0\.0|NaN|NaN|NaN|
|
| 54 |
+
|
| 55 |
# Intended Use
|
| 56 |
- Real-time detections on underwater footage
|
| 57 |
- Post-processed video/imagery for detecting sea urchins in underwater environments
|