def __init__(self, cfg):
super(AnyNet, self).__init__()
self.cfg = cfg.copy()
self.max_disp = cfg.model.max_disp
self.stage = cfg.model.stage
self.backbone = build_backbone(cfg)
self.cost_processor = build_cost_processor(cfg)
# disparity predictor
self.disp_predictor = nn.ModuleDict()
for st in self.stage:
self.disp_predictor[st] = FasterSoftArgmin(
max_disp=cfg.model.disp_predictor.max_disp[st],
start_disp=cfg.model.disp_predictor.start_disp[st],
dilation=cfg.model.disp_predictor.dilation[st],
alpha=cfg.model.disp_predictor.alpha,
normalize=cfg.model.disp_predictor.normalize,
)
self.disp_refinement = build_disp_refinement(cfg)
# make general stereo matching loss evaluator
self.loss_evaluator = make_gsm_loss_evaluator(cfg)
def setUpClass(cls):
cls.device = torch.device('cuda:7')
config_path = '/home/zhixiang/youmin/projects/depth/public/' \
'DenseMatchingBenchmark/configs/AnyNet/scene_flow.py'
cls.cfg = Config.fromfile(config_path)
cls.backbone = build_backbone(cls.cfg)
cls.backbone.to(cls.device)
cls.setUpTimeTestingClass()
cls.avg_time = {}
def __init__(self, cfg):
super(DeepPruner, self).__init__()
self.cfg = cfg.copy()
self.max_disp = cfg.model.max_disp
self.backbone = build_backbone(cfg)
self.disp_sampler = build_disp_sampler(cfg)
self.cost_processor = build_cost_processor(cfg)
self.disp_refinement = build_disp_refinement(cfg)
# make general stereo matching loss evaluator
self.loss_evaluator = make_gsm_loss_evaluator(cfg)
def __init__(self, cfg):
super(GeneralizedStereoModel, self).__init__()
self.cfg = cfg.copy()
self.max_disp = cfg.model.max_disp
self.scale = cfg.model.backbone.scale
self.backbone = build_backbone(cfg)
self.cost_processor = build_cost_processor(cfg)
# confidence measurement network
self.cmn = None
if 'cmn' in cfg.model:
self.cmn = build_cmn(cfg)
self.disp_predictor = build_disp_predictor(cfg)
self.loss_evaluator = make_gsm_loss_evaluator(cfg)