Hugging Face's logo

monai-test
/
pathology_nuclei_classification

monai
medical
Model card Files
xet
Community
katielink commited on
Commit
2349c14
·
1 Parent(s): 2c47f4e

update benchmark on A100

Browse files
Files changed (7) hide show
  1. README.md +13 -13
  2. configs/metadata.json +4 -3
  3. configs/multi_gpu_evaluate.json +1 -1
  4. configs/multi_gpu_train.json +1 -1
  5. docs/README.md +13 -13
  6. models/model.pt +1 -1
  7. models/model.ts +2 -2
README.md CHANGED
@@ -21,7 +21,7 @@ The training dataset is from https://warwick.ac.uk/fac/cross_fac/tia/data/hovern
21
  wget https://warwick.ac.uk/fac/cross_fac/tia/data/hovernet/consep_dataset.zip
22
  unzip -q consep_dataset.zip
23
  ```
24
- ![](images/dataset.jpeg)<br/>
25
 
26
  ## Training configuration
27
  The training was performed with the following:
@@ -103,22 +103,22 @@ Example `dataset.json` in output folder:
103
  - 2 = Epithelial
104
  - 3 = Spindle-Shaped
105
 
106
- ![](images/val_in_out.jpeg)
107
 
108
  ## Scores
109
  This model achieves the following F1 score on the validation data provided as part of the dataset:
110
 
111
- - Train F1 score = 0.96
112
- - Validation F1 score = 0.85
113
 
114
  <hr/>
115
  Confusion Metrics for <b>Validation</b> for individual classes are (at epoch 50):
116
 
117
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
118
  |-----------|--------|--------------|------------|----------------|
119
- | Precision | 0.5846 | 0.7143 | 0.9158 | 0.8399 |
120
- | Recall | 0.2550 | 0.8441 | 0.9193 | 0.8106 |
121
- | F1-score | 0.3551 | 0.7738 | 0.9175 | 0.8250 |
122
 
123
 
124
  <hr/>
@@ -126,22 +126,22 @@ Confusion Metrics for <b>Training</b> for individual classes are (at epoch 50):
126
 
127
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
128
  |-----------|--------|--------------|------------|----------------|
129
- | Precision | 0.9059 | 0.9569 | 0.9754 | 0.9494 |
130
- | Recall | 0.8370 | 0.9547 | 0.9790 | 0.9502 |
131
- | F1-score | 0.8701 | 0.9558 | 0.9772 | 0.9498 |
132
 
133
 
134
 
135
  ## Training Performance
136
  A graph showing the training Loss and F1-score over 50 epochs.
137
 
138
- ![](images/train_loss.jpeg) <br>
139
- ![](images/train_f1.jpeg) <br>
140
 
141
  ## Validation Performance
142
  A graph showing the validation F1-score over 50 epochs.
143
 
144
- ![](images/val_f1.jpeg) <br>
145
 
146
 
147
  ## commands example
 
21
  wget https://warwick.ac.uk/fac/cross_fac/tia/data/hovernet/consep_dataset.zip
22
  unzip -q consep_dataset.zip
23
  ```
24
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_dataset.jpeg)<br/>
25
 
26
  ## Training configuration
27
  The training was performed with the following:
 
103
  - 2 = Epithelial
104
  - 3 = Spindle-Shaped
105
 
106
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_val_in_out.jpeg)
107
 
108
  ## Scores
109
  This model achieves the following F1 score on the validation data provided as part of the dataset:
110
 
111
+ - Train F1 score = 0.941
112
+ - Validation F1 score = 0.840
113
 
114
  <hr/>
115
  Confusion Metrics for <b>Validation</b> for individual classes are (at epoch 50):
116
 
117
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
118
  |-----------|--------|--------------|------------|----------------|
119
+ | Precision | 0.6250 | 0.7085 | 0.9188 | 0.8571 |
120
+ | Recall | 0.1449 | 0.8750 | 0.9310 | 0.8154 |
121
+ | F1-score | 0.2353 | 0.7830 | 0.9249 | 0.8357 |
122
 
123
 
124
  <hr/>
 
126
 
127
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
128
  |-----------|--------|--------------|------------|----------------|
129
+ | Precision | 0.8902 | 0.9418 | 0.9717 | 0.9189 |
130
+ | Recall | 0.7935 | 0.9250 | 0.9725 | 0.9345 |
131
+ | F1-score | 0.8391 | 0.9333 | 0.9721 | 0.9267 |
132
 
133
 
134
 
135
  ## Training Performance
136
  A graph showing the training Loss and F1-score over 50 epochs.
137
 
138
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_loss_v2.png) <br>
139
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_f1_v2.png) <br>
140
 
141
  ## Validation Performance
142
  A graph showing the validation F1-score over 50 epochs.
143
 
144
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_val_f1_v2.png) <br>
145
 
146
 
147
  ## commands example
configs/metadata.json CHANGED
@@ -1,7 +1,8 @@
1
  {
2
  "schema": "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20220324.json",
3
- "version": "0.0.6",
4
  "changelog": {
 
5
  "0.0.6": "adapt to BundleWorkflow interface",
6
  "0.0.5": "add name tag",
7
  "0.0.4": "Fix evaluation",
@@ -9,7 +10,7 @@
9
  "0.0.2": "Update The Torch Vision Transform",
10
  "0.0.1": "initialize the model package structure"
11
  },
12
- "monai_version": "1.2.0rc3",
13
  "pytorch_version": "1.13.1",
14
  "numpy_version": "1.22.2",
15
  "optional_packages_version": {
@@ -28,7 +29,7 @@
28
  "label_classes": "single channel data",
29
  "pred_classes": "4 channels OneHot data, channel 0 is Other, channel 1 is Inflammatory, channel 2 is Epithelial, channel 3 is Spindle-Shaped",
30
  "eval_metrics": {
31
- "f1_score": 0.85
32
  },
33
  "intended_use": "This is an example, not to be used for diagnostic purposes",
34
  "references": [
 
1
  {
2
  "schema": "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20220324.json",
3
+ "version": "0.0.7",
4
  "changelog": {
5
+ "0.0.7": "update benchmark on A100",
6
  "0.0.6": "adapt to BundleWorkflow interface",
7
  "0.0.5": "add name tag",
8
  "0.0.4": "Fix evaluation",
 
10
  "0.0.2": "Update The Torch Vision Transform",
11
  "0.0.1": "initialize the model package structure"
12
  },
13
+ "monai_version": "1.2.0rc4",
14
  "pytorch_version": "1.13.1",
15
  "numpy_version": "1.22.2",
16
  "optional_packages_version": {
 
29
  "label_classes": "single channel data",
30
  "pred_classes": "4 channels OneHot data, channel 0 is Other, channel 1 is Inflammatory, channel 2 is Epithelial, channel 3 is Spindle-Shaped",
31
  "eval_metrics": {
32
+ "f1_score": 0.84
33
  },
34
  "intended_use": "This is an example, not to be used for diagnostic purposes",
35
  "references": [
configs/multi_gpu_evaluate.json CHANGED
@@ -31,6 +31,6 @@
31
  "$@validate#evaluator.run()"
32
  ],
33
  "finalize": [
34
- "$dist.destroy_process_group()"
35
  ]
36
  }
 
31
  "$@validate#evaluator.run()"
32
  ],
33
  "finalize": [
34
+ "$dist.is_initialized() and dist.destroy_process_group()"
35
  ]
36
  }
configs/multi_gpu_train.json CHANGED
@@ -40,6 +40,6 @@
40
  "$@train#trainer.run()"
41
  ],
42
  "finalize": [
43
- "$dist.destroy_process_group()"
44
  ]
45
  }
 
40
  "$@train#trainer.run()"
41
  ],
42
  "finalize": [
43
+ "$dist.is_initialized() and dist.destroy_process_group()"
44
  ]
45
  }
docs/README.md CHANGED
@@ -14,7 +14,7 @@ The training dataset is from https://warwick.ac.uk/fac/cross_fac/tia/data/hovern
14
  wget https://warwick.ac.uk/fac/cross_fac/tia/data/hovernet/consep_dataset.zip
15
  unzip -q consep_dataset.zip
16
  ```
17
- ![](images/dataset.jpeg)<br/>
18
 
19
  ## Training configuration
20
  The training was performed with the following:
@@ -96,22 +96,22 @@ Example `dataset.json` in output folder:
96
  - 2 = Epithelial
97
  - 3 = Spindle-Shaped
98
 
99
- ![](images/val_in_out.jpeg)
100
 
101
  ## Scores
102
  This model achieves the following F1 score on the validation data provided as part of the dataset:
103
 
104
- - Train F1 score = 0.96
105
- - Validation F1 score = 0.85
106
 
107
  <hr/>
108
  Confusion Metrics for <b>Validation</b> for individual classes are (at epoch 50):
109
 
110
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
111
  |-----------|--------|--------------|------------|----------------|
112
- | Precision | 0.5846 | 0.7143 | 0.9158 | 0.8399 |
113
- | Recall | 0.2550 | 0.8441 | 0.9193 | 0.8106 |
114
- | F1-score | 0.3551 | 0.7738 | 0.9175 | 0.8250 |
115
 
116
 
117
  <hr/>
@@ -119,22 +119,22 @@ Confusion Metrics for <b>Training</b> for individual classes are (at epoch 50):
119
 
120
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
121
  |-----------|--------|--------------|------------|----------------|
122
- | Precision | 0.9059 | 0.9569 | 0.9754 | 0.9494 |
123
- | Recall | 0.8370 | 0.9547 | 0.9790 | 0.9502 |
124
- | F1-score | 0.8701 | 0.9558 | 0.9772 | 0.9498 |
125
 
126
 
127
 
128
  ## Training Performance
129
  A graph showing the training Loss and F1-score over 50 epochs.
130
 
131
- ![](images/train_loss.jpeg) <br>
132
- ![](images/train_f1.jpeg) <br>
133
 
134
  ## Validation Performance
135
  A graph showing the validation F1-score over 50 epochs.
136
 
137
- ![](images/val_f1.jpeg) <br>
138
 
139
 
140
  ## commands example
 
14
  wget https://warwick.ac.uk/fac/cross_fac/tia/data/hovernet/consep_dataset.zip
15
  unzip -q consep_dataset.zip
16
  ```
17
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_dataset.jpeg)<br/>
18
 
19
  ## Training configuration
20
  The training was performed with the following:
 
96
  - 2 = Epithelial
97
  - 3 = Spindle-Shaped
98
 
99
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_val_in_out.jpeg)
100
 
101
  ## Scores
102
  This model achieves the following F1 score on the validation data provided as part of the dataset:
103
 
104
+ - Train F1 score = 0.941
105
+ - Validation F1 score = 0.840
106
 
107
  <hr/>
108
  Confusion Metrics for <b>Validation</b> for individual classes are (at epoch 50):
109
 
110
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
111
  |-----------|--------|--------------|------------|----------------|
112
+ | Precision | 0.6250 | 0.7085 | 0.9188 | 0.8571 |
113
+ | Recall | 0.1449 | 0.8750 | 0.9310 | 0.8154 |
114
+ | F1-score | 0.2353 | 0.7830 | 0.9249 | 0.8357 |
115
 
116
 
117
  <hr/>
 
119
 
120
  | Metric | Other | Inflammatory | Epithelial | Spindle-Shaped |
121
  |-----------|--------|--------------|------------|----------------|
122
+ | Precision | 0.8902 | 0.9418 | 0.9717 | 0.9189 |
123
+ | Recall | 0.7935 | 0.9250 | 0.9725 | 0.9345 |
124
+ | F1-score | 0.8391 | 0.9333 | 0.9721 | 0.9267 |
125
 
126
 
127
 
128
  ## Training Performance
129
  A graph showing the training Loss and F1-score over 50 epochs.
130
 
131
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_loss_v2.png) <br>
132
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_train_f1_v2.png) <br>
133
 
134
  ## Validation Performance
135
  A graph showing the validation F1-score over 50 epochs.
136
 
137
+ ![](https://developer.download.nvidia.com/assets/Clara/Images/monai_pathology_classification_val_f1_v2.png) <br>
138
 
139
 
140
  ## commands example
models/model.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f254ae6b0318e1375d48c1e9d6056d236d4a1a32957afc4aeafba0e047c46b2b
3
  size 28419489
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9aef3d7c3114f1c86695e119d64f7e83dcf6449716cb64df9d5b7b447adb8a5
3
  size 28419489
models/model.ts CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a3d6a7da0c74a470cb2e9cbd72c5783ec0f2c7ed120cc6dfada30338b73637dd
3
- size 28575181
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37d348addc879bc51d5e5fa89927ceb48c3521632135a0d5cdc488592961209e
3
+ size 28620879