Edmond Yoo
2018-10-13 23d94e4846bf4ec13069703a28b1d776f4bbe44f
opencv_dnn.py
@@ -9,8 +9,17 @@
from PIL import Image
import time
from config import Config
import fetch_data
import transform_data
"""
As of the current version, the YOLO network has been removed from this code during optimization.
It was found out that YOLO was adding too much processing delay, and the benefits from using it couldn't justify
such heavy cost.
If you're interested to see the implementation using YOLO, please check out the previous commit:
https://github.com/hj3yoo/mtg_card_detector/tree/dea64611730c84a59c711c61f7f80948f82bcd31
"""
def calc_image_hashes(card_pool, save_to=None, hash_size=32, highfreq_factor=4):
@@ -45,7 +54,7 @@
        for card_name in card_names:
            # Fetch the image - name can be found based on the card's information
            card_info['name'] = card_name
            img_name = '%s/card_img/png/%s/%s_%s.png' % (transform_data.data_dir, card_info['set'],
            img_name = '%s/card_img/png/%s/%s_%s.png' % (Config.data_dir, card_info['set'],
                                                         card_info['collector_number'],
                                                         fetch_data.get_valid_filename(card_info['name']))
            card_img = cv2.imread(img_name)
@@ -53,7 +62,7 @@
            # If the image doesn't exist, download it from the URL
            if card_img is None:
                fetch_data.fetch_card_image(card_info,
                                            out_dir='%s/card_img/png/%s' % (transform_data.data_dir, card_info['set']))
                                            out_dir='%s/card_img/png/%s' % (Config.data_dir, card_info['set']))
                card_img = cv2.imread(img_name)
            if card_img is None:
                print('WARNING: card %s is not found!' % img_name)
@@ -275,7 +284,7 @@
    # Find the contour
    _, cnts, hier = cv2.findContours(img_erode, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    if len(cnts) == 0:
        print('no contours')
        #print('no contours')
        return []
    # The hierarchy from cv2.findContours() is similar to a tree: each node has an access to the parent, the first child
@@ -334,7 +343,7 @@
        card_set = key[key.find('(') + 1:key.find(')')]
        confidence = sum(val) / f_len
        card_info = card_pool[(card_pool['name'] == card_name) & (card_pool['set'] == card_set)].iloc[0]
        img_name = '%s/card_img/tiny/%s/%s_%s.png' % (transform_data.data_dir, card_info['set'],
        img_name = '%s/card_img/tiny/%s/%s_%s.png' % (Config.data_dir, card_info['set'],
                                                      card_info['collector_number'],
                                                      fetch_data.get_valid_filename(card_info['name']))
        # If the card image is not found, just leave it blank
@@ -533,8 +542,8 @@
    else:
        # Merge database for all cards, then calculate pHash values of each, store them
        df_list = []
        for set_name in fetch_data.all_set_list:
            csv_name = '%s/csv/%s.csv' % (transform_data.data_dir, set_name)
        for set_name in Config.all_set_list:
            csv_name = '%s/csv/%s.csv' % (Config.data_dir, set_name)
            df = fetch_data.load_all_cards_text(csv_name)
            df_list.append(df)
        card_pool = pd.concat(df_list, sort=True)