Exemplo n.º 1
0
    def test_sr_predict_shape(self, datamodel):
        DataModel = datamodel
        src = DataModel.g.dataset_uri("__data__", None)
        dst = DataModel.g.dataset_uri("001_gaussian_blur", group="features")
        result = survos.run_command(
            "features", "gaussian_blur", uri=None, src=src, dst=dst
        )

        with DatasetManager(src, out=dst, dtype="float32", fillvalue=0) as DM:
            src_dataset = DM.sources[0]
            dst_dataset = DM.out
            src_arr = src_dataset[:]
            gblur_arr = dst_dataset[:]

        result = survos.run_command("superregions", "create", uri=None)
        features_src = DataModel.g.dataset_uri("001_gaussian_blur", group="features")
        dst = DataModel.g.dataset_uri("001_superregions", group="superregions")

        result = supervoxels(
            features_src,
            dst,
            n_segments=8,
            compactness=0.5,
            spacing=[1, 1, 1],
            multichannel=False,
            enforce_connectivity=False,
        )
        with DatasetManager(src, out=dst, dtype="float32", fillvalue=0) as DM:
            src_dataset = DM.sources[0]
            dst_dataset = DM.out
            src_arr = src_dataset[:]
            dst_arr = dst_dataset[:]

        superseg_cfg = cfg.pipeline
        superseg_cfg["type"] = "rf"
        superseg_cfg["predict_params"]["clf"] = "Ensemble"

        refine = False
        lam = (1.0,)

        anno_arr = np.ones_like(dst_arr)
        anno_arr[2:4, 2:4, 2:4] = 2
        feature_arr = view_dataset("001_gaussian_blur", "features", 3)
        segmentation = sr_predict(
            dst_arr,
            anno_arr,
            [feature_arr, gblur_arr],
            None,
            superseg_cfg,
            refine,
            lam,
        )
Exemplo n.º 2
0
def run_workflow(workflow_file):
    if not os.path.isabs(workflow_file):
        workflow_file = os.path.join(os.getcwd(), workflow_file)

        with open(workflow_file) as f:
            workflows = yaml.safe_load(f.read())

        num_workflow_steps = len(workflows.keys())
        minVal, maxVal = 0, num_workflow_steps

        print(workflows)

        for step_idx, k in enumerate(workflows):
            workflow = workflows[k]
            action = workflow.pop("action")
            plugin, command = action.split(".")
            params = workflow.pop("params")

            src_name = workflow.pop("src")
            dst_name = workflow.pop("dst")

            if "src_group" in workflow:
                plugin = workflow.pop("src_group")

            src = DataModel.g.dataset_uri(src_name, group=plugin)
            dst = DataModel.g.dataset_uri(dst_name, group=plugin)

            all_params = dict(src=src, dst=dst, modal=True)
            all_params.update(params)
            logger.info(f"Executing workflow {all_params}")

            print(
                f"+ Running {k}, with {plugin}, {command} on {src}\n to dst {dst} {all_params}\n"
            )

            import survos

            # Launcher.g.run(plugin, command, **all_params)
            survos.run_command(plugin, command, uri=None, src=src, dst=dst)

            # src_arr = view_dataset(dst_name, plugin, 10)

    else:
        print("Need input workflow YAML file")

    return all_params, params
Exemplo n.º 3
0
    def test_feature_generation(self, datamodel):
        DataModel = datamodel

        src = DataModel.g.dataset_uri("__data__", None)
        with DatasetManager(src, out=None, dtype="float32", fillvalue=0) as DM:
            src_dataset = DM.sources[0]
            raw_arr = src_dataset[:]

        random_blobs = binary_blobs(length=max(raw_arr.shape), n_dim=3)
        random_blobs_anno = np.zeros_like(raw_arr)
        random_blobs_anno[
            0 : raw_arr.shape[0], 0 : raw_arr.shape[1], 0 : raw_arr.shape[2]
        ] = random_blobs[
            0 : raw_arr.shape[0], 0 : raw_arr.shape[1], 0 : raw_arr.shape[2]
        ]
        result = survos.run_command(
            "annotations",
            "add_level",
            uri=None,
            workspace=DataModel.g.current_workspace,
        )
        assert "id" in result[0]

        level_id = result[0]["id"]
        label_values = np.unique(random_blobs_anno)

        for v in label_values:
            params = dict(
                level=level_id,
                idx=int(v),
                name=str(v),
                color="#11FF11",
                workspace=DataModel.g.current_workspace,
            )
        label_result = survos.run_command("annotations", "add_label", **params)

        dst = DataModel.g.dataset_uri(level_id, group="annotations")
        with DatasetManager(dst, out=dst, dtype="uint32", fillvalue=0) as DM:
            DM.out[:] = random_blobs_anno
Exemplo n.º 4
0
    def test_superregions_shape(self, datamodel):
        DataModel = datamodel
        src = DataModel.g.dataset_uri("__data__", None)
        dst = DataModel.g.dataset_uri("001_gaussian_blur", group="features")
        result = survos.run_command(
            "features", "gaussian_blur", uri=None, src=src, dst=dst
        )
        result = survos.run_command(
            "superregions", "create", uri=None, workspace=DataModel.g.current_workspace
        )
        assert len(result) == 2
        assert result[0]["kind"] == "supervoxels"
        features_src = DataModel.g.dataset_uri("001_gaussian_blur", group="features")
        dst = DataModel.g.dataset_uri("001_superregions", group="superregions")

        n_segments = 8
        result = supervoxels(
            features_src,
            dst,
            n_segments=n_segments,
            compactness=1,
            spacing=[1, 1, 1],
            multichannel=False,
            enforce_connectivity=False,
        )

        assert result["n_segments"] == n_segments

        with DatasetManager(src, out=dst, dtype="float32", fillvalue=0) as DM:
            src_dataset = DM.sources[0]
            dst_dataset = DM.out
            src_arr = src_dataset[:]
            dst_arr = dst_dataset[:]

        assert dst_arr.shape == src_arr.shape
        assert len(np.unique(dst_arr)) == n_segments
Exemplo n.º 5
0
def datamodel():
    # make test vol
    map_fullpath = os.path.join("./tmp/testvol_4x4x4b.h5")

    testvol = np.array(
        [
            [
                [0.1761602, 0.6701295, 0.13151232, 0.95726678],
                [0.4795476, 0.48114134, 0.0410548, 0.29893265],
                [0.49127266, 0.70298447, 0.42751211, 0.08101552],
                [0.73805652, 0.83111601, 0.36852477, 0.38732476],
            ],
            [
                [0.2847222, 0.96054574, 0.25430756, 0.35403861],
                [0.54439093, 0.65897414, 0.1959487, 0.90714872],
                [0.84462152, 0.90754182, 0.02455657, 0.26180662],
                [0.1711208, 0.40122666, 0.54562598, 0.01419861],
            ],
            [
                [0.59280376, 0.42706895, 0.86637913, 0.87831645],
                [0.57991401, 0.31989204, 0.85869799, 0.6333411],
                [0.21539274, 0.63780214, 0.64204493, 0.74425482],
                [0.1903691, 0.81962537, 0.31774673, 0.34812628],
            ],
            [
                [0.40880077, 0.595773, 0.28856063, 0.19316746],
                [0.03195766, 0.62475541, 0.50762591, 0.34700798],
                [0.98913461, 0.07883111, 0.96534233, 0.57697606],
                [0.71496714, 0.70764578, 0.92294417, 0.91300531],
            ],
        ]
    )
    # testvol = np.ones((4,4,4)).astype(np.float32) / 2.0
    # print(testvol)

    with h5py.File(map_fullpath, "w") as hf:
        hf.create_dataset("data", data=testvol)

    print(DataModel.g.CHROOT)
    tmp_ws_name = "testworkspace_tmp2"

    result = survos.run_command("workspace", "get", uri=None, workspace=tmp_ws_name)

    if not type(result[0]) == dict:
        logger.debug("Creating temp workspace")
        survos.run_command("workspace", "create", uri=None, workspace=tmp_ws_name)
    else:
        logger.debug("tmp exists, deleting and recreating")
        survos.run_command("workspace", "delete", uri=None, workspace=tmp_ws_name)
        logger.debug("workspace deleted")
        survos.run_command("workspace", "create", uri=None, workspace=tmp_ws_name)
        logger.debug("workspace recreated")

    # add data to workspace
    survos.run_command(
        "workspace",
        "add_data",
        uri=None,
        workspace=tmp_ws_name,
        data_fname=map_fullpath,
        dtype="float32",
    )

    DataModel.g.current_workspace = tmp_ws_name

    return DataModel