add workflow, train loss and validation accuracy figures

README.md

@@ -12,6 +12,8 @@ A pre-trained model for the endoscopic inbody classification task.
 This model is trained using the SEResNet50 structure, whose details can be found in [1]. All datasets are from private samples of [Activ Surgical](https://www.activsurgical.com/). Samples in training and validation dataset are from the same 4 videos, while test samples are from different two videos.
 The [pytorch model](https://drive.google.com/file/d/14CS-s1uv2q6WedYQGeFbZeEWIkoyNa-x/view?usp=sharing) and [torchscript model](https://drive.google.com/file/d/1fOoJ4n5DWKHrt9QXTZ2sXwr9C-YvVGCM/view?usp=sharing) are shared in google drive. Modify the `bundle_root` parameter specified in `configs/train.json` and `configs/inference.json` to reflect where models are downloaded. Expected directory path to place downloaded models is `models/` under `bundle_root`.
 
+![image](https://developer.download.nvidia.com/assets/Clara/Images/monai_endoscopic_inbody_classification_workflow.png)
+
 ## Data
 Datasets used in this work were provided by [Activ Surgical](https://www.activsurgical.com/). Here is a [link](https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/inbody_outbody_samples.zip) of 20 samples (10 in-body and 10 out-body) to show what this dataset looks like. After downloading this dataset, python script in `scripts` folder naming `data_process` can be used to get label json files by running the command below and replacing datapath and outpath parameters.
 ```
@@ -59,6 +61,16 @@ This model achieves the following accuracy score on the test dataset:
 
 Accuracy = 0.98
 
+## Training Performance
+A graph showing the training loss over 25 epochs.
+
+![](https://developer.download.nvidia.com/assets/Clara/Images/monai_endoscopic_inbody_classification_train_loss.png) <br>
+
+## Validation Performance
+A graph showing the validation accuracy over 25 epochs.
+
+![](https://developer.download.nvidia.com/assets/Clara/Images/monai_endoscopic_inbody_classification_val_accuracy.png) <br>
+
 ## commands example
 Execute training:
 
@@ -109,23 +121,6 @@ python -m monai.bundle ckpt_export network_def \
     --config_file configs/inference.json
 ```
 
-Export checkpoint to onnx file, which has been tested on pytorch 1.12.0:
-
-```
-python scripts/export_to_onnx.py --model models/model.pt --outpath models/model.onnx
-```
-
-Export TensorRT float16 model from the onnx model:
-
-```
-trtexec --onnx=models/model.onnx --saveEngine=models/model.trt --fp16 \
-    --minShapes=INPUT__0:1x3x256x256 \
-    --optShapes=INPUT__0:16x3x256x256 \
-    --maxShapes=INPUT__0:32x3x256x256 \
-    --shapes=INPUT__0:8x3x256x256
-```
-This command need TensorRT with correct CUDA installed in the environment. For the detail of installing TensorRT, please refer to [this link](https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html).
-
 # References
 [1] J. Hu, L. Shen and G. Sun, Squeeze-and-Excitation Networks, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018, pp. 7132-7141. https://arxiv.org/pdf/1709.01507.pdf
 

configs/metadata.json

@@ -1,7 +1,8 @@
 {
     "schema": "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20220324.json",
-    "version": "0.3.0",
+    "version": "0.3.1",
     "changelog": {
+        "0.3.1": "add workflow, train loss and validation accuracy figures",
         "0.3.0": "update dataset processing",
         "0.2.2": "update to use monai 1.0.1",
         "0.2.1": "enhance readme on commands example",

docs/README.md

@@ -5,6 +5,8 @@ A pre-trained model for the endoscopic inbody classification task.
 This model is trained using the SEResNet50 structure, whose details can be found in [1]. All datasets are from private samples of [Activ Surgical](https://www.activsurgical.com/). Samples in training and validation dataset are from the same 4 videos, while test samples are from different two videos.
 The [pytorch model](https://drive.google.com/file/d/14CS-s1uv2q6WedYQGeFbZeEWIkoyNa-x/view?usp=sharing) and [torchscript model](https://drive.google.com/file/d/1fOoJ4n5DWKHrt9QXTZ2sXwr9C-YvVGCM/view?usp=sharing) are shared in google drive. Modify the `bundle_root` parameter specified in `configs/train.json` and `configs/inference.json` to reflect where models are downloaded. Expected directory path to place downloaded models is `models/` under `bundle_root`.
 
+![image](https://developer.download.nvidia.com/assets/Clara/Images/monai_endoscopic_inbody_classification_workflow.png)
+
 ## Data
 Datasets used in this work were provided by [Activ Surgical](https://www.activsurgical.com/). Here is a [link](https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/inbody_outbody_samples.zip) of 20 samples (10 in-body and 10 out-body) to show what this dataset looks like. After downloading this dataset, python script in `scripts` folder naming `data_process` can be used to get label json files by running the command below and replacing datapath and outpath parameters.
 ```
@@ -52,6 +54,16 @@ This model achieves the following accuracy score on the test dataset:
 
 Accuracy = 0.98
 
+## Training Performance
+A graph showing the training loss over 25 epochs.
+
+![](https://developer.download.nvidia.com/assets/Clara/Images/monai_endoscopic_inbody_classification_train_loss.png) <br>
+
+## Validation Performance
+A graph showing the validation accuracy over 25 epochs.
+
+![](https://developer.download.nvidia.com/assets/Clara/Images/monai_endoscopic_inbody_classification_val_accuracy.png) <br>
+
 ## commands example
 Execute training:
 
@@ -102,23 +114,6 @@ python -m monai.bundle ckpt_export network_def \
     --config_file configs/inference.json
 ```
 
-Export checkpoint to onnx file, which has been tested on pytorch 1.12.0:
-
-```
-python scripts/export_to_onnx.py --model models/model.pt --outpath models/model.onnx
-```
-
-Export TensorRT float16 model from the onnx model:
-
-```
-trtexec --onnx=models/model.onnx --saveEngine=models/model.trt --fp16 \
-    --minShapes=INPUT__0:1x3x256x256 \
-    --optShapes=INPUT__0:16x3x256x256 \
-    --maxShapes=INPUT__0:32x3x256x256 \
-    --shapes=INPUT__0:8x3x256x256
-```
-This command need TensorRT with correct CUDA installed in the environment. For the detail of installing TensorRT, please refer to [this link](https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html).
-
 # References
 [1] J. Hu, L. Shen and G. Sun, Squeeze-and-Excitation Networks, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018, pp. 7132-7141. https://arxiv.org/pdf/1709.01507.pdf