env collection, smaller changes

README.md

@@ -18,6 +18,32 @@ The resulting self-configuring method, nnDetection, adapts itself without any ma
 We demonstrate the effectiveness of nnDetection on two public benchmarks, ADAM and LUNA16, and propose 10 further public data sets for a comprehensive evaluation of medical object detection methods.
 
 # Installation
+## Docker Installation
+The easiest way to get started with nnDetection is the provided is to build a Docker Container with the provided Dockerfile.
+
+Please install docker and [nvidia-docker2](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html) before continuing.
+
+All projects which are based on nnDetection assume that the base image was built with the following tagging scheme `nnDetection:[version]`.
+To build a container (nnDetection Version 0.1) run the following command from the base directory:
+
+```bash
+docker build -t nndetection:0.1 --build-arg env_det_num_threads=6 --build-arg env_det_verbose=1 .
+```
+
+(`--build-arg env_det_num_threads=6` and `--build-arg env_det_verbose=1` are optional and are used to overwrite the provided default parameters)
+
+The docker container expects data and models in its own `/opt/data` and `/opt/models` directories respectively.
+The directories need to be mounted via docker `-v`. For simplicity and speed, the ENV variables `det_data` and `det_models` can be set in the host system to point to the desired directories. To run:
+
+```bash
+docker run --gpus all -v ${det_data}:/opt/data -v ${det_models}:/opt/models -it --shm-size=24gb nndetection:0.1 /bin/bash
+```
+
+Warning:
+When running a training inside the container it is necessary to [increase the shared memory](https://stackoverflow.com/questions/30210362/how-to-increase-the-size-of-the-dev-shm-in-docker-container) (via --shm-size).
+
+
+## Source Installation
 1. Install CUDA (>10.1) and cudnn (make sure to select [compatible versions](https://docs.nvidia.com/deeplearning/cudnn/support-matrix/index.html)!)
 2. [Optional] Depending on your GPU you might need to set `TORCH_CUDA_ARCH_LIST`, check [compute capabilities](https://developer.nvidia.com/cuda-gpus) here.
 3. Install [torch](https://pytorch.org/) (make sure to match the pytorch and CUDA versions!) (requires pytorch >1.7+)
@@ -52,34 +78,6 @@ To test the whole installation please run the Toy Dataset example.
 To get the best possible performance we recommend using CUDA 11.0+ with cuDNN 8.1.X+ and a (!)locally compiled version(!) of Pytorch 1.7+
 </details>
 
-<details close>
-<summary>Docker Container</summary>
-<br>
-The provided Dockerfile can be used to setup quick development environments or deploy nnDetection.
-
-Please install docker and [nvidia-docker2](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html) before continuing.
-
-All projects which are based on nnDetection assume that the base image was built with the following tagging scheme `nnDetection:[version]`.
-To build a container (nnDetection Version 0.1) run the following command from the base directory:
-
-```bash
-docker build -t nndetection:0.1 --build-arg env_det_num_threads=6 --build-arg env_det_verbose=1 .
-```
-
-(`--build-arg env_det_num_threads=6` and `--build-arg env_det_verbose=1` are optional and are used to overwrite the provided default parameters)
-
-The docker container expects data and models in its own `/opt/data` and `/opt/models` directories respectively.
-The directories need to be mounted via docker `-v`. For simplicity and speed, the ENV variables `det_data` and `det_models` can be set in the host system to point to the desired directories. To run:
-
-```bash
-docker run --gpus all -v ${det_data}:/opt/data -v ${det_models}:/opt/models -it --shm-size=24gb nndetection:0.1 /bin/bash
-```
-
-Warning:
-When running a training inside the container it is necessary to [increase the shared memory](https://stackoverflow.com/questions/30210362/how-to-increase-the-size-of-the-dev-shm-in-docker-container) (via --shm-size).
-
-</details>
-
 # nnDetection
 <div align="center">
     <img src=docs/source/nnDetectionFunctional.svg width="600px">
@@ -134,11 +132,12 @@ After running the generation script follow the `Planning`, `Training` and `Infer
 </div>
 
 ## Adding New Datasets
-nnDetection relie on a standardized input format which is very similar to the [nnU-Net](https://github.com/MIC-DKFZ/nnUNet) format and allows easy integration of new datasets.
-The format is explained below.
+nnDetection relies on a standardized input format which is very similar to the [nnU-Net](https://github.com/MIC-DKFZ/nnUNet) format and allows easy integration of new datasets.
+More details about the format can be found below.
 
 ### Folders
 All datasets should reside inside `Task[Number]_[Name]` folder inside the specified detection data folder (et the path to this folder with the `det_data` environment flag).
+To avoid conflicts with our provided pretrained models we recommend to use task numbers starting from 100.
 An overview is provided below ([Name] symbolise folder, `-` symbolise files, indents refer to substructures)
 
 ```text
@@ -317,7 +316,7 @@ nndet_eval 000 RetinaUNetV001_D3V001_3d 0 --boxes --analyze_boxes
 ```
 
 ### Inference
-After running all fold it is time to collect the models and creat a unified inference plan.
+After running all folds it is time to collect the models and creat a unified inference plan.
 The following command will copy all the models and predictions per fold and by adding the `sweep` options the empiricaly hyperparameter optimization across all fold can be started.
 This will generate a unified plan for all models which will be used during inference.
 
@@ -331,6 +330,7 @@ nndet_consolidate 000 RetinaUNetV001_D3V001_3d --sweep_boxes
 # /experiments/consolidate.py - main()
 ```
 
+For the final test set predictions simply select the best model according to the validation scores and run the prediction command below.
 Data which is located in `raw_splitted/imagesTs` will be automatically preprocessed and predicted by running the following command:
 ```bash
 nndet_predict [task] [model] [--fold] [--num_models] [--num_tta] [--no_preprocess]

nndet/planning/properties/instance.py

@@ -155,7 +155,7 @@ def analyze_instances_per_case(analyzer: DatasetAnalyzer,
     props["num_instances"] = count_instances(props, all_classes)
     props["has_classes"] = list(set(props["instances"].values()))
     props["volume_per_class"], props["region_volume_per_class"] = \
-        instance_class_and_region_sizes(iseg, props, all_classes)
+        instance_class_and_region_sizes(case_id, iseg, props, all_classes)
     props["boxes"] = iseg_to_boxes(iseg)
     props["all_ious"], props["class_ious"] = case_ious(props["boxes"], props)
     return props
@@ -178,6 +178,7 @@ def count_instances(props: dict, all_classes: Sequence[int]) -> Dict[int, int]:
 
 
 def instance_class_and_region_sizes(
+        case_id: str,
         iseg: np.ndarray,
         props: dict,
         all_classes: Sequence[int],
@@ -209,7 +210,8 @@ def instance_class_and_region_sizes(
     ids = np.unique(iseg)
     ids = ids[ids > 0]
     if len(ids) != len(list(instance_classes.keys())):
-        logger.warning(f"Instance lost. Found {instance_classes} in properties but {ids} in seg.")
+        logger.warning(f"Instance lost. Found {instance_classes} in "
+                       f"properties but {ids} in seg of {case_id}.")
     volumer_per_instance = {c: np.sum(iseg == c) * vol_per_voxel for c in ids}
 
     for instance_id, instance_vol in volumer_per_instance.items():
@@ -278,13 +280,10 @@ def case_ious(boxes: np.ndarray, props: dict) -> Tuple[np.ndarray, Dict[int, np.
         class_ious = OrderedDict()
         case_classes = list(set(map(int, props["instances"].values())))
         case_instances = sorted(list(map(int, props["instances"].keys())))
-        try:  # TODO: fix bug in case of missing instances
-            for cls in case_classes:
-                cls_box_indices = [props["instances"][str(ci)] == cls for ci in case_instances]
-                class_ious[cls] = compute_each_iou(boxes[cls_box_indices])
-        except:
-            pass
-            # from IPython import embed; embed()
+
+        for cls in case_classes:
+            cls_box_indices = [props["instances"][str(ci)] == cls for ci in case_instances]
+            class_ious[cls] = compute_each_iou(boxes[cls_box_indices])
     else:
         all_ious = np.array([])
         class_ious = {}

scripts/utils.py

@@ -178,3 +178,21 @@ def unpack():
     p = args.path
     num_processes = args.num_processes
     unpack_dataset(p, num_processes, False)
+
+
+def env():
+    import os
+    import torch
+    print(f"PyTorch Version: {torch.version}")
+    print(f"PyTorch CUDA: {torch.version.cuda}")
+    print(f"PyTorch Backend cudnn: {torch.backends.cudnn.version()}")
+    print(f"PyTorch CUDA Arch List: {torch.cuda.get_arch_list()}")
+    print(f"PyTorch Current Device Capability: {torch.cuda.get_device_capability()}")
+    print(f"PyTorch CUDA available: {torch.cuda.is_available()}")
+
+    stream = os.popen('nvcc --version')
+    output = stream.read()
+    print(f"System NVCC: {output}")
+    print(f"System Arch List: {os.getenv('TORCH_CUDA_ARCH_LIST', None)}")
+    print(f"System OMP_NUM_THREADS: {os.getenv('OMP_NUM_THREADS', None)}")
+    print(f"System CUDA_HOME is None: {os.getenv('CUDA_HOME', None) is None}")

setup.py

@@ -128,6 +128,7 @@ setup(
             'nndet_boxes2nii = scripts.utils:boxes2nii',
             'nndet_seg2nii = scripts.utils:seg2nii',
             'nndet_unpack = scripts.utils:unpack',
+            'nndet_env = scripts.utils:env',
         ]
     },
 )