Python cfg.DATA_DIR属性代码示例

# 需要导入模块: from core.config import cfg [as 别名]
# 或者: from core.config.cfg import DATA_DIR [as 别名]
def parse_args():
    """Parser command line argumnets"""
    parser = argparse.ArgumentParser(formatter_class=ColorHelpFormatter)
    parser.add_argument('--output_dir', help='Directory to save downloaded weight files',
                        default=os.path.join(cfg.DATA_DIR, 'pretrained_model'))
    parser.add_argument('-t', '--targets', nargs='+', metavar='file_name',
                        help='Files to download. Allowed values are: ' +
                        ', '.join(map(lambda s: Fore.YELLOW + s + Fore.RESET,
                                      list(PRETRAINED_WEIGHTS.keys()))),
                        choices=list(PRETRAINED_WEIGHTS.keys()),
                        default=list(PRETRAINED_WEIGHTS.keys()))
    return parser.parse_args()


# ---------------------------------------------------------------------------- #
# Mapping from filename to google drive file_id
# ---------------------------------------------------------------------------- # 

# 需要导入模块: from core.config import cfg [as 别名]
# 或者: from core.config.cfg import DATA_DIR [as 别名]
def cache_path(self):
        cache_path = os.path.abspath(os.path.join(cfg.DATA_DIR, 'cache'))
        if not os.path.exists(cache_path):
            os.makedirs(cache_path)
        return cache_path 

# 需要导入模块: from core.config import cfg [as 别名]
# 或者: from core.config.cfg import DATA_DIR [as 别名]
def cache_path(self):
        cache_path = os.path.abspath(os.path.join(cfg.DATA_DIR, 'cache'))
        if cfg.TRAIN.GT_SCORES:
            cache_path += '_gt-scores'
        if cfg.TRAIN.JOINT_TRAINING:
            cache_path += '_joint'
        if not os.path.exists(cache_path):
            os.makedirs(cache_path)
        return cache_path 

# 需要导入模块: from core.config import cfg [as 别名]
# 或者: from core.config.cfg import DATA_DIR [as 别名]
def get_rel_counts(ds_name, must_overlap=True):
    """
    Get counts of all of the relations. Used for modeling directly P(rel | o1, o2)
    :param train_data: 
    :param must_overlap: 
    :return: 
    """

    if ds_name.find('vg') >= 0:
        with open(cfg.DATA_DIR + '/vg/rel_annotations_train.json') as f:
            train_data = json.load(f)
    elif ds_name.find('vrd') >= 0:
        with open(cfg.DATA_DIR + '/vrd/new_annotations_train.json') as f:
            train_data = json.load(f)
    else:
        raise NotImplementedError

    fg_matrix = np.zeros((
        cfg.MODEL.NUM_CLASSES - 1,  # not include background
        cfg.MODEL.NUM_CLASSES - 1,  # not include background
        cfg.MODEL.NUM_PRD_CLASSES + 1,  # include background
    ), dtype=np.int64)

    bg_matrix = np.zeros((
        cfg.MODEL.NUM_CLASSES - 1,  # not include background
        cfg.MODEL.NUM_CLASSES - 1,  # not include background
    ), dtype=np.int64)
    
    for _, im_rels in train_data.items():
        # get all object boxes
        gt_box_to_label = {}
        for i, rel in enumerate(im_rels):
            sbj_box = box_utils.y1y2x1x2_to_x1y1x2y2(rel['subject']['bbox'])
            obj_box = box_utils.y1y2x1x2_to_x1y1x2y2(rel['object']['bbox'])
            sbj_lbl = rel['subject']['category']  # not include background
            obj_lbl = rel['object']['category']  # not include background
            prd_lbl = rel['predicate']  # not include background
            if tuple(sbj_box) not in gt_box_to_label:
                gt_box_to_label[tuple(sbj_box)] = sbj_lbl
            if tuple(obj_box) not in gt_box_to_label:
                gt_box_to_label[tuple(obj_box)] = obj_lbl
            
            fg_matrix[sbj_lbl, obj_lbl, prd_lbl + 1] += 1
        
        if cfg.MODEL.USE_OVLP_FILTER:
            if len(gt_box_to_label):
                gt_boxes = np.array(list(gt_box_to_label.keys()), dtype=np.int32)
                gt_classes = np.array(list(gt_box_to_label.values()), dtype=np.int32)
                o1o2_total = gt_classes[np.array(
                    box_filter(gt_boxes, must_overlap=must_overlap), dtype=int)]
                for (o1, o2) in o1o2_total:
                    bg_matrix[o1, o2] += 1
        else:
            # consider all pairs of boxes, overlapped or non-overlapped
            for b1, l1 in gt_box_to_label.items():
                for b2, l2 in gt_box_to_label.items():
                    if b1 == b2:
                        continue
                    bg_matrix[l1, l2] += 1

    return fg_matrix, bg_matrix 

# 需要导入模块: from core.config import cfg [as 别名]
# 或者: from core.config.cfg import DATA_DIR [as 别名]
def get_obj_prd_vecs(dataset_name):
    word2vec_model = gensim.models.KeyedVectors.load_word2vec_format(
        cfg.DATA_DIR + '/word2vec_model/GoogleNews-vectors-negative300.bin', binary=True)
    logger.info('Model loaded.')
    # change everything into lowercase
    all_keys = list(word2vec_model.vocab.keys())
    for key in all_keys:
        new_key = key.lower()
        word2vec_model.vocab[new_key] = word2vec_model.vocab.pop(key)
    logger.info('Wiki words converted to lowercase.')

    if dataset_name.find('vrd') >= 0:
        with open(cfg.DATA_DIR + '/vrd/objects.json') as f:
            obj_cats = json.load(f)
        with open(cfg.DATA_DIR + '/vrd/predicates.json') as f:
            prd_cats = json.load(f)
    elif dataset_name.find('vg') >= 0:
        with open(cfg.DATA_DIR + '/vg/objects.json') as f:
            obj_cats = json.load(f)
        with open(cfg.DATA_DIR + '/vg/predicates.json') as f:
            prd_cats = json.load(f)
    else:
        raise NotImplementedError
    # represent background with the word 'unknown'
    # obj_cats.insert(0, 'unknown')
    prd_cats.insert(0, 'unknown')
    all_obj_vecs = np.zeros((len(obj_cats), 300), dtype=np.float32)
    for r, obj_cat in enumerate(obj_cats):
        obj_words = obj_cat.split()
        for word in obj_words:
            raw_vec = word2vec_model[word]
            all_obj_vecs[r] += (raw_vec / la.norm(raw_vec))
        all_obj_vecs[r] /= len(obj_words)
    logger.info('Object label vectors loaded.')
    all_prd_vecs = np.zeros((len(prd_cats), 300), dtype=np.float32)
    for r, prd_cat in enumerate(prd_cats):
        prd_words = prd_cat.split()
        for word in prd_words:
            raw_vec = word2vec_model[word]
            all_prd_vecs[r] += (raw_vec / la.norm(raw_vec))
        all_prd_vecs[r] /= len(prd_words)
    logger.info('Predicate label vectors loaded.')
    return all_obj_vecs, all_prd_vecs