Example #1
0
    def get_valloader(self):
        if self.configer.get('method') == 'single_shot_detector':
            valloader = data.DataLoader(
                SSDDataLoader(root_dir=os.path.join(
                    self.configer.get('data', 'data_dir'), 'val'),
                              aug_transform=self.aug_val_transform,
                              img_transform=self.img_transform,
                              configer=self.configer),
                batch_size=self.configer.get('data', 'val_batch_size'),
                shuffle=False,
                num_workers=self.configer.get('data', 'workers'),
                pin_memory=True)

            return valloader

        elif self.configer.get('method') == 'faster_rcnn':
            valloader = data.DataLoader(
                FRDataLoader(root_dir=os.path.join(
                    self.configer.get('data', 'data_dir'), 'val'),
                             aug_transform=self.aug_val_transform,
                             img_transform=self.img_transform,
                             configer=self.configer),
                batch_size=self.configer.get('data', 'val_batch_size'),
                shuffle=False,
                num_workers=self.configer.get('data', 'workers'),
                pin_memory=True)

            return valloader

        else:
            Log.error('Method: {} loader is invalid.'.format(
                self.configer.get('method')))
            return None
Example #2
0
    def get_trainloader(self):
        if self.configer.get('method') == 'single_shot_detector':
            trainloader = data.DataLoader(
                SSDDataLoader(root_dir=os.path.join(
                    self.configer.get('data', 'data_dir'), 'train'),
                              aug_transform=self.aug_train_transform,
                              img_transform=self.img_transform,
                              configer=self.configer),
                batch_size=self.configer.get('train', 'batch_size'),
                shuffle=True,
                num_workers=self.configer.get('data', 'workers'),
                pin_memory=True,
                collate_fn=lambda *args: CollateFunctions.our_collate(
                    *args,
                    data_keys=['img', 'bboxes', 'labels'],
                    configer=self.configer,
                    trans_dict=self.configer.get('train', 'data_transformer')))

            return trainloader

        elif self.configer.get('method') == 'faster_rcnn':
            trainloader = data.DataLoader(
                FRDataLoader(root_dir=os.path.join(
                    self.configer.get('data', 'data_dir'), 'train'),
                             aug_transform=self.aug_train_transform,
                             img_transform=self.img_transform,
                             configer=self.configer),
                batch_size=self.configer.get('train', 'batch_size'),
                shuffle=True,
                num_workers=self.configer.get('data', 'workers'),
                pin_memory=True,
                collate_fn=lambda *args: CollateFunctions.our_collate(
                    *args,
                    data_keys=['img', 'imgscale', 'bboxes', 'labels'],
                    configer=self.configer,
                    trans_dict=self.configer.get('train', 'data_transformer')))

            return trainloader

        elif self.configer.get('method') == 'yolov3':
            trainloader = data.DataLoader(
                YOLODataLoader(root_dir=os.path.join(
                    self.configer.get('data', 'data_dir'), 'train'),
                               aug_transform=self.aug_train_transform,
                               img_transform=self.img_transform,
                               configer=self.configer),
                batch_size=self.configer.get('train', 'batch_size'),
                shuffle=True,
                num_workers=self.configer.get('data', 'workers'),
                pin_memory=True,
                collate_fn=lambda *args: CollateFunctions.our_collate(
                    *args,
                    data_keys=['img', 'bboxes', 'labels'],
                    configer=self.configer,
                    trans_dict=self.configer.get('train', 'data_transformer')))

            return trainloader

        else:
            Log.error('Method: {} loader is invalid.'.format(
                self.configer.get('method')))
            return None