code
stringlengths
17
6.64M
def main(): parser = argparse.ArgumentParser(description='Convert keys in official pretrained segformer to MMSegmentation style.') parser.add_argument('src', help='src model path or url') parser.add_argument('dst', help='save path') args = parser.parse_args() checkpoint = CheckpointLoader.load_che...
def convert_vit(ckpt): new_ckpt = OrderedDict() for (k, v) in ckpt.items(): if k.startswith('head'): continue if k.startswith('norm'): new_k = k.replace('norm.', 'ln1.') elif k.startswith('patch_embed'): if ('proj' in k): new_k = k.re...
def main(): parser = argparse.ArgumentParser(description='Convert keys in timm pretrained vit models to MMSegmentation style.') parser.add_argument('src', help='src model path or url') parser.add_argument('dst', help='save path') args = parser.parse_args() checkpoint = CheckpointLoader.load_checkp...
def parse_args(): parser = argparse.ArgumentParser(description='Print the whole config') parser.add_argument('config', help='config file path') parser.add_argument('--graph', action='store_true', help='print the models graph') parser.add_argument('--options', nargs='+', action=DictAction, help='argume...
def main(): args = parse_args() cfg = Config.fromfile(args.config) if (args.options is not None): cfg.merge_from_dict(args.options) print(f'''Config: {cfg.pretty_text}''') cfg.dump('example.py') if args.graph: model = init_segmentor(args.config, device='cpu') print(f'''...
def parse_args(): parser = argparse.ArgumentParser(description='Process a checkpoint to be published') parser.add_argument('in_file', help='input checkpoint filename') parser.add_argument('out_file', help='output checkpoint filename') args = parser.parse_args() return args
def process_checkpoint(in_file, out_file): checkpoint = torch.load(in_file, map_location='cpu') if ('optimizer' in checkpoint): del checkpoint['optimizer'] torch.save(checkpoint, out_file) sha = subprocess.check_output(['sha256sum', out_file]).decode() final_file = (out_file.rstrip('.pth')...
def main(): args = parse_args() process_checkpoint(args.in_file, args.out_file)
def digit_version(version_str): digit_version = [] for x in version_str.split('.'): if x.isdigit(): digit_version.append(int(x)) elif (x.find('rc') != (- 1)): patch_version = x.split('rc') digit_version.append((int(patch_version[0]) - 1)) digit_v...
def check_torch_version(): torch_minimum_version = '1.8.0' torch_version = digit_version(torch.__version__) assert (torch_version >= digit_version(torch_minimum_version)), f'Torch=={torch.__version__} is not support for converting to torchscript. Please install pytorch>={torch_minimum_version}.'
def _convert_batchnorm(module): module_output = module if isinstance(module, torch.nn.SyncBatchNorm): module_output = torch.nn.BatchNorm2d(module.num_features, module.eps, module.momentum, module.affine, module.track_running_stats) if module.affine: module_output.weight.data = modu...
def _demo_mm_inputs(input_shape, num_classes): 'Create a superset of inputs needed to run test or train batches.\n\n Args:\n input_shape (tuple):\n input batch dimensions\n num_classes (int):\n number of semantic classes\n ' (N, C, H, W) = input_shape rng = np.ran...
def pytorch2libtorch(model, input_shape, show=False, output_file='tmp.pt', verify=False): 'Export Pytorch model to TorchScript model and verify the outputs are\n same between Pytorch and TorchScript.\n\n Args:\n model (nn.Module): Pytorch model we want to export.\n input_shape (tuple): Use thi...
def parse_args(): parser = argparse.ArgumentParser(description='Convert MMSeg to TorchScript') parser.add_argument('config', help='test config file path') parser.add_argument('--checkpoint', help='checkpoint file', default=None) parser.add_argument('--show', action='store_true', help='show TorchScript...
def parse_args(): parser = argparse.ArgumentParser(description='mmseg test (and eval) a model') parser.add_argument('config', help='test config file path') parser.add_argument('checkpoint', help='checkpoint file') parser.add_argument('--work-dir', help='if specified, the evaluation metric results will...
def main(): args = parse_args() assert (args.out or args.eval or args.format_only or args.show or args.show_dir), 'Please specify at least one operation (save/eval/format/show the results / save the results) with the argument "--out", "--eval", "--format-only", "--show" or "--show-dir"' if (args.eval and ...
def mmseg2torchserve(config_file: str, checkpoint_file: str, output_folder: str, model_name: str, model_version: str='1.0', force: bool=False): "Converts mmsegmentation model (config + checkpoint) to TorchServe\n `.mar`.\n\n Args:\n config_file:\n In MMSegmentation config format.\n ...
def parse_args(): parser = ArgumentParser(description='Convert mmseg models to TorchServe `.mar` format.') parser.add_argument('config', type=str, help='config file path') parser.add_argument('checkpoint', type=str, help='checkpoint file path') parser.add_argument('--output-folder', type=str, required...
class MMsegHandler(BaseHandler): def initialize(self, context): properties = context.system_properties self.map_location = ('cuda' if torch.cuda.is_available() else 'cpu') self.device = torch.device((((self.map_location + ':') + str(properties.get('gpu_id'))) if torch.cuda.is_available() ...
def parse_args(): parser = ArgumentParser(description='Compare result of torchserve and pytorch,and visualize them.') parser.add_argument('img', help='Image file') parser.add_argument('config', help='Config file') parser.add_argument('checkpoint', help='Checkpoint file') parser.add_argument('model...
def main(args): url = ((('http://' + args.inference_addr) + '/predictions/') + args.model_name) with open(args.img, 'rb') as image: tmp_res = requests.post(url, image) content = tmp_res.content if args.result_image: with open(args.result_image, 'wb') as out_image: out_image...
def parse_args(): parser = argparse.ArgumentParser(description='Train a segmentor') parser.add_argument('config', help='train config file path') parser.add_argument('--work-dir', help='the dir to save logs and models') parser.add_argument('--load-from', help='the checkpoint file to load weights from')...
def main(): args = parse_args() cfg = Config.fromfile(args.config) if (args.options is not None): cfg.merge_from_dict(args.options) if cfg.get('cudnn_benchmark', False): torch.backends.cudnn.benchmark = True if (args.work_dir is not None): cfg.work_dir = args.work_dir e...
def parse_args(): parser = argparse.ArgumentParser(description='Train a segmentor') parser.add_argument('config', help='train config file path') parser.add_argument('--work-dir', help='the dir to save logs and models') parser.add_argument('--load-from', help='the checkpoint file to load weights from')...
def main(): args = parse_args() cfg = Config.fromfile(args.config) if (args.options is not None): cfg.merge_from_dict(args.options) if cfg.get('cudnn_benchmark', False): torch.backends.cudnn.benchmark = True if (args.work_dir is not None): cfg.work_dir = args.work_dir e...
def _parse_args(): (args_config, remaining) = config_parser.parse_known_args() if args_config.config: with open(args_config.config, 'r') as f: cfg = yaml.safe_load(f) parser.set_defaults(**cfg) args = parser.parse_args(remaining) args_text = yaml.safe_dump(args.__dict__...
def main(): setup_default_logging() (args, args_text) = _parse_args() if args.log_wandb: if has_wandb: wandb.init(project=args.experiment, config=args) else: _logger.warning("You've requested to log metrics to wandb but package not found. Metrics not being logged to...
def train_one_epoch(epoch, model, loader, optimizer, loss_fn, args, lr_scheduler=None, saver=None, output_dir=None, amp_autocast=suppress, loss_scaler=None, model_ema=None, mixup_fn=None): if (args.mixup_off_epoch and (epoch >= args.mixup_off_epoch)): if (args.prefetcher and loader.mixup_enabled): ...
def validate(model, loader, loss_fn, args, amp_autocast=suppress, log_suffix=''): batch_time_m = AverageMeter() losses_m = AverageMeter() top1_m = AverageMeter() top5_m = AverageMeter() model.eval() end = time.time() last_idx = (len(loader) - 1) with torch.no_grad(): for (batch...
def get_command_args(): parser = argparse.ArgumentParser() parser.add_argument('--dataname', '-n', help='name of generated', default='sewfactory') parser.add_argument('--config', '-c', help='config file for dataset resource', default='meta_infos\\configs\\data_sim_configs.json') parser.add_argument('-...
def init_mayapy(): try: print('Initilializing Maya tools...') maya.standalone.initialize() print('Load plugins') cmds.loadPlugin('mtoa.mll') cmds.loadPlugin('objExport.mll') cmds.loadPlugin('fbxmaya.mll') except Exception as e: print(e) pass
def stop_mayapy(): maya.standalone.uninitialize() print('Maya stopped')
def get_command_args(): parser = argparse.ArgumentParser() parser.add_argument('--dataname', '-n', help='name of generated', default='deepfashion') parser.add_argument('--config', '-c', help='config file for dataset resource', default='meta_infos\\configs\\data_sim_configs.json') parser.add_argument('...
def init_mayapy(): try: print('Initilializing Maya tools...') maya.standalone.initialize() print('Load plugins') cmds.loadPlugin('mtoa.mll') cmds.loadPlugin('objExport.mll') cmds.loadPlugin('fbxmaya.mll') except Exception as e: print(e) pass
def stop_mayapy(): maya.standalone.uninitialize() print('Maya stopped')
def get_command_args(): 'command line arguments to control the run' parser = argparse.ArgumentParser() parser.add_argument('--base-config', '-c', help='template config with parameters used for animation', default='meta_infos\\configs\\anime_config.json') parser.add_argument('--base-fbx', '-f', help='i...
def _create_data_folder(path, props): ' Create a new directory to put dataset in \n & generate appropriate name & update dataset properties\n ' if ('data_folder' in props): props['name'] = (props['data_folder'] + '_regen') data_folder = props['name'] else: data_folder = P...
def generate(path, templates_path, props): 'Generates a synthetic dataset of patterns with given properties\n Params:\n path : path to folder to put a new dataset into\n templates_path : path to folder with pattern templates\n props : an instance of DatasetProperties class\...
def get_command_args(): 'command line arguments to control the run' parser = argparse.ArgumentParser() parser.add_argument('--config', '-c', help='pattern config', type=str, default='meta_infos/configs/dataset_config.yaml') parser.add_argument('--out', '-o', help='folder to save generated patterns', t...
class Properties(): 'Keeps, loads, and saves cofiguration & statistic information\n Supports gets&sets as a dictionary\n Provides shortcuts for batch-init configurations\n\n One of the usages -- store system-dependent basic cofiguration\n ' def __init__(self, filename='', clean_stats=...
class Garment(object): def __init__(self, name, type): self.name = name self.type = type def to_filter_string(self): return ((self.type + '/') + self.name) def to_rel_folder(self): return os.path.join(self.type, self.name) def to_abs_path(self, data_root): r...
class GarmentMaterials(object): '\n Describes the materials for rendering.\n mtl_scene: a scene file contains all the supportted mtls\n\n Note: support 4 kinds of mtls now (default, cotton, velvet, silk).\n Need input mtl resources first. \n Assumes: the code is based ...
class MayaScene(object): '\n Decribes scene setup that includes:\n # Mtl(s) & light(s): preload, donot move\n * floor & camera(s): preload, ajdust according to the body\n Assumes \n * body the scene revolved aroung faces z+ direction\n ' def __init__(self, pr...
class GarmentPlayblast(object): '\n 1. load scene\n 2. load smplbody \n 3. load garment\n 4. run simulation\n 5. run render\n ' def __init__(self, conf): self.config = conf self.default_center = [0.037, (- 29.154), 2.363] self.smooth_cameras = {} ...
class PredictPlayblast(GarmentPlayblast): def __init__(self, conf): super(PredictPlayblast, self).__init__(conf) self.set_panel_render_camera() def set_panel_render_camera(self): self.panel_cameras = {} (cam, camshape) = cmds.camera(aspectRatio=1, name='panel_front') ...
def load_plugin(): '\n Forces loading Qualoth plugin into Maya. \n Note that plugin should be installed and licensed to use it!\n Inquire here: http://www.fxgear.net/vfxpricing\n ' maya_year = int(mel.eval('getApplicationVersionAsFloat')) plugin_name = (('qualoth_' + str(maya_year)...
def qlCreatePattern(curves_group): '\n Converts given 2D closed curve to a flat geometry piece\n ' objects_before = cmds.ls(assemblies=True) cmds.select(curves_group) mel.eval('qlCreatePattern()') objects_after = cmds.ls(assemblies=True) patterns = list((set(objects_after) - set(obje...
def qlCreateSeam(curve1, curve2): '\n Create a seam between two selected curves\n ' cmds.select([curve1, curve2]) seam_shape = mel.eval('qlCreateSeam()') return seam_shape
def qlCreateCollider(cloth, target): "\n Marks object as a collider object for cloth --\n eshures that cloth won't penetrate body when simulated\n " objects_before = cmds.ls(assemblies=True) cmds.select([cloth, target]) mel.eval('qlCreateCollider()') objects_after = cmds.ls(assemb...
def qlCreateAttachConstraint(points, target): objects_before = cmds.ls(assemblies=True) if (not isinstance(target, list)): cmds.select((points + [target])) else: cmds.select((points + target)) mel.eval('qlCreateAttachConstraint()') objects_after = cmds.ls(assemblies=True) const...
def qlCleanCache(cloth): 'Clean layback cache for given cloth. Accepts qlCloth object' cmds.select(cloth) mel.eval('qlClearCache()')
def qlReinitSolver(cloth, solver): 'Reinitialize solver \n set both cloth and solver to the initial state before simulation was applied\n NOTE: useful for correct reload of garments on delete\n ' cmds.select([cloth, solver]) mel.eval('qlReinitializeSolver()')
def start_maya_sim(garment, props): 'Start simulation through Maya defalut playback without checks\n Gives Maya user default control over stopping & resuming sim\n Current qlCloth material properties from Maya are used (instead of garment config)\n ' config = props['config'] solver = _ini...
def run_sim(garment, props): '\n Setup and run cloth simulator untill static equlibrium is achieved.\n Note:\n * Assumes garment is already properly aligned!\n * All of the garments existing in Maya scene will be simulated\n because solver is shared!!\n ' ...
def findSolver(): '\n Returns the name of the qlSover existing in the scene\n (usully solver is created once per scene)\n ' solver = cmds.ls('*qlSolver*Shape*') return (solver[0] if solver else None)
def deleteSolver(): 'deletes all solver objects from the scene' cmds.delete(cmds.ls('*qlSolver*'))
def flipPanelNormal(panel_geom): 'Set flippling normals to True for a given panel geom objects\n at least one of the provided objects should a qlPattern object' ql_pattern = [obj for obj in panel_geom if ('Pattern' in obj)] ql_pattern = ql_pattern[0] shape = cmds.listRelatives(ql_pattern, shape...
def getVertsOnCurve(panel_node, curve, curve_group=None): "\n Return the list of mesh vertices located on the curve \n * panel_node is qlPattern object to which the curve belongs\n * curve is a main name of a curve object to get vertex info for\n OR any substring of it's full Maya ...
def setColliderFriction(collider_objects, friction_value): 'Sets the level of friction of the given collider to friction_value' main_collider = [obj for obj in collider_objects if ('Offset' not in obj)] collider_shape = cmds.listRelatives(main_collider[0], shapes=True) cmds.setAttr((collider_shape[0] ...
def setFabricProps(cloth, props): 'Set given material propertied to qlClothObject' if (not props): return cmds.setAttr((cloth + '.density'), props['density'], clamp=True) cmds.setAttr((cloth + '.stretch'), props['stretch_resistance'], clamp=True) cmds.setAttr((cloth + '.shear'), props['she...
def setPanelsResolution(scaling): 'Set resoluiton conroller of all qlPatterns in the scene' all_panels = cmds.ls('*qlPattern*', shapes=True) for panel in all_panels: cmds.setAttr((panel + '.resolutionScale'), scaling)
def fetchFabricProps(cloth): "Returns current material properties of the cloth's objects\n Requires qlCloth object\n " props = {} props['density'] = cmds.getAttr((cloth + '.density')) props['stretch_resistance'] = cmds.getAttr((cloth + '.stretch')) props['shear_resistance'] = cmds.getAtt...
def fetchColliderFriction(collider_objects): 'Retrieve collider friction info from given collider' try: main_collider = [obj for obj in collider_objects if ('Offset' not in obj)] collider_shape = cmds.listRelatives(main_collider[0], shapes=True) return cmds.getAttr((collider_shape[0] +...
def fetchPanelResolution(): some_panels = cmds.ls('*qlPattern*') shape = cmds.listRelatives(some_panels[0], shapes=True, path=True) return cmds.getAttr((shape[0] + '.resolutionScale'))
def _init_sim(config): '\n Basic simulation settings before starting simulation\n ' solver = findSolver() cmds.setAttr((solver + '.selfCollision'), 1) cmds.setAttr((solver + '.startTime'), 1) cmds.setAttr((solver + '.solverStatistics'), 0) cmds.playbackOptions(ps=0, max=config['max_s...
def _set_gravity(solver, gravity): 'Set a given value of gravity to sim solver' cmds.setAttr((solver + '.gravity1'), gravity)
def _update_progress(progress, total): 'Progress bar in console' amtDone = (progress / total) num_dash = int((amtDone * 50)) sys.stdout.write('\rProgress: [{0:50s}] {1:.1f}%'.format((('#' * num_dash) + ('-' * (50 - num_dash))), (amtDone * 100))) sys.stdout.flush()
def _record_fail(props, fail_type, garment_name): "add a failure recording to props. Creates nodes if don't exist" if ('fails' not in props['stats']): props['stats']['fails'] = {} try: props['stats']['fails'][fail_type].append(garment_name) except KeyError: props['stats']['fail...
def single_file_sim(resources, props, caching=False): '\n Simulates the given template and puts the results in original template folder, \n including config and statistics\n ' try: init_sim_props(props, True) qw.load_plugin() scene = mymaya.Scene(os.path.join(resources...
def batch_sim(resources, data_path, dataset_props, num_samples=None, caching=False, force_restart=False): '\n Performs pattern simulation for each example in the dataset \n given by dataset_props. \n Batch processing is automatically resumed \n from the last unporcessed datapoint if re...
def batch_sim_with_mtls(resources, data_path, dataset_props, mtls=None, num_samples=None, caching=False, force_restart=False): if (('frozen' in dataset_props) and dataset_props['frozen']): print('Warning: dataset is frozen, processing is skipped') return True resume = init_sim_props(dataset_pr...
def init_sim_props(props, batch_run=False, force_restart=False): ' \n Add default config values if not given in props & clean-up stats if not resuming previous processing\n Returns a flag wheter current simulation is a resumed last one\n ' if ('sim' not in props): props.set_section_co...
def template_simulation(spec, scene, sim_props, delete_on_clean=False, caching=False, save_maya_scene=False): '\n Simulate given template within given scene & save log files\n ' print('\nGarment load') garment = mymaya.MayaGarment(spec) try: garment.load(shader_group=scene.cloth_SG()...
def template_simulation_with_mtls(spec, scene, sim_props, delete_on_clean=False, caching=False, save_maya_scene=False): shd_names = list(scene.Mtls.material_types) num_body = 1 sim_names = list(sim_props.keys()) names = list(set(shd_names).intersection(sim_names)) for name in names: print(...
def _serialize_props_with_sim_stats(dataset_props, filename): 'Compute data processing statistics and serialize props to file' dataset_props.stats_summary() dataset_props.serialize(filename)
def _get_pattern_files(data_path, dataset_props): ' Collects paths to all the pattern files in given folder' to_ignore = ['renders'] pattern_specs = [] (root, dirs, files) = next(os.walk(data_path)) if dataset_props['to_subfolders']: for directory in dirs: if (directory not in ...
class NumpyArrayEncoder(JSONEncoder): def default(self, obj): if isinstance(obj, np.ndarray): return obj.tolist() return JSONEncoder.default(self, obj)
def load_file(filepath, name='object'): 'Load mesh to the scene' if (not os.path.isfile(filepath)): raise RuntimeError('Loading Object from file to Maya::Missing file {}'.format(filepath)) obj = cmds.file(filepath, i=True, rnn=True)[0] obj = cmds.rename(obj, (name + '#')) return obj
def save_mesh(target, to_file): 'Save given object to file as a mesh' cmds.select(clear=True) cmds.select(target) cmds.file(to_file, type='OBJExport', exportSelectedStrict=True, options='groups=0;ptgroups=0;materials=0;smoothing=0;normals=1', force=True, defaultExtensions=False) cmds.select(clear=...
def get_dag(object_name): 'Return DAG for requested object' selectionList = OpenMaya.MSelectionList() selectionList.add(object_name) dag = OpenMaya.MDagPath() selectionList.getDagPath(0, dag) return dag
def get_mesh_dag(object_name): 'Return MFnMesh object by the object name' dag = get_dag(object_name) mesh = OpenMaya.MFnMesh(dag) return (mesh, dag)
def get_vertices_np(mesh): '\n Retreive vertex info as np array for given mesh object\n ' maya_vertices = OpenMaya.MPointArray() mesh.getPoints(maya_vertices, OpenMaya.MSpace.kWorld) vertices = np.empty((maya_vertices.length(), 3)) for i in range(maya_vertices.length()): for j in...
def match_vert_lists(short_list, long_list): '\n Find the vertices from long list that correspond to verts in short_list\n Both lists are numpy arrays\n NOTE: Assuming order is matching => O(len(long_list)) complexity: \n order of vertices in short list is the same as in long list ...
def test_ray_intersect(mesh, raySource, rayVector, accelerator=None, hit_tol=None, return_info=False): 'Check if given ray intersect given mesh\n * hit_tol ignores intersections that are within hit_tol from the ray source (as % of ray length) -- usefull when checking self-intersect\n * mesh is expec...
def edge_vert_ids(mesh, edge_id): 'Return vertex ids for a given edge in given mesh' script_util = OpenMaya.MScriptUtil(0.0) v_ids_cptr = script_util.asInt2Ptr() mesh.getEdgeVertices(edge_id, v_ids_cptr) ty = (ctypes.c_uint * 2) v_ids_list = ty.from_address(int(v_ids_cptr)) return (v_ids_l...
def scale_to_cm(target, max_height_cm=220): 'Heuristically check the target units and scale to cantimeters if other units are detected\n * default value of max_height_cm is for meshes of humans\n ' bb = cmds.polyEvaluate(target, boundingBox=True) height = (bb[1][1] - bb[1][0]) if (height < (...
def eulerAngleToRoatationMatrix(theta): R_x = np.array([[1, 0, 0], [0, math.cos(theta[0]), (- math.sin(theta[0]))], [0, math.sin(theta[0]), math.cos(theta[0])]]) R_y = np.array([[math.cos(theta[1]), 0, math.sin(theta[1])], [0, 1, 0], [(- math.sin(theta[1])), 0, math.cos(theta[1])]]) R_z = np.array([[math....
def isRotationMatrix(R): Rt = np.transpose(R) shouldBeIdentity = np.dot(Rt, R) I = np.identity(3, dtype=R.dtype) n = np.linalg.norm((I - shouldBeIdentity)) return (n < 1e-06)
def rotationMatrixToEulerAngles(R): assert isRotationMatrix(R) sy = math.sqrt(((R[(0, 0)] * R[(0, 0)]) + (R[(1, 0)] * R[(1, 0)]))) singular = (sy < 1e-06) if (not singular): x = math.atan2(R[(2, 1)], R[(2, 2)]) y = math.atan2((- R[(2, 0)]), sy) z = math.atan2(R[(1, 0)], R[(0, 0...
def load_pose_data(data_file): spin = (True if data_file.endswith('.json') else False) if spin: data = json.load(open(data_file, 'r')) if ('rotmat_tuned' in data): rotmat = np.array(data['rotmat_tuned']) else: rotmat = np.array(data['rotmat']) poses = []...
def _Rx(theta): return np.matrix([[1, 0, 0], [0, m.cos(theta), (- m.sin(theta))], [0, m.sin(theta), m.cos(theta)]])
def _Ry(theta): return np.matrix([[m.cos(theta), 0, m.sin(theta)], [0, 1, 0], [(- m.sin(theta)), 0, m.cos(theta)]])
def _Rz(theta): return np.matrix([[m.cos(theta), (- m.sin(theta)), 0], [m.sin(theta), m.cos(theta), 0], [0, 0, 1]])
def euler_xyz_to_R(euler): 'Convert to Rotation matrix.\n Expects input in degrees.\n Only support Maya convension of intrinsic xyz Euler Angles\n ' return ((_Rz(np.deg2rad(euler[2])) * _Ry(np.deg2rad(euler[1]))) * _Rx(np.deg2rad(euler[0])))
def R_to_euler(R): '\n Convert Rotation matrix to Euler-angles in degrees (in Maya convension of intrinsic xyz Euler Angles)\n NOTE: \n Routine produces one of the possible Euler angles, corresponding to input rotations (the Euler angles are not uniquely defined)\n ' tol = (sys.flo...
def copy2cpu(tensor): if isinstance(tensor, np.ndarray): return tensor return tensor.detach().cpu().numpy()
class PanelClasses(): ' Interface to access panel classification by role ' def __init__(self, classes_file): self.filename = classes_file with open(classes_file, 'r') as f: self.classes = json.load(f, object_pairs_hook=OrderedDict) self.names = list(self.classes.keys()) ...
def flip_img(img): 'Flip rgb images or masks.\n channels come last, e.g. (256,256,3).\n ' img = np.fliplr(img) return img
def _dict_to_tensors(dict_obj): 'convert a dictionary with numeric values into a new dictionary with torch tensors' new_dict = dict.fromkeys(dict_obj.keys()) for (key, value) in dict_obj.items(): if (key == 'image'): new_dict[key] = value elif (value is None): new_d...
class SampleToTensor(object): 'Convert ndarrays in sample to Tensors.' def __call__(self, sample): return _dict_to_tensors(sample)