[ { "path": "readme.md", "content": "# PokerVision\nRecognition of all entities on the poker table( now only for Pokerstars) and added analytics on the basis of which you can make decisions about your moves \n\n![pokerstars.gif](pokerstars.gif)\n\n\nNOTE: it only works with the theme that I have and for 6 players. It has been tested on ubuntu with python version 3.8\n\n\n# Installing and Running\n\n```\n$ git clone https://github.com/wb-08/PokerVision\n$ cd PokerVision\n$ virtualenv venv\n$ source venv/bin/activate\n$ pip3 install -r requirements.txt\n$ cd scripts\n$ python3 grab_table.py\n```\n" }, { "path": "requirements.txt", "content": "eval7==0.1.9\nfuture==0.18.3\nmss==9.0.1\nnumpy==1.24.3\nopencv-contrib-python==4.5.5.62\nopencv-python==4.5.5.62\nPillow==9.5.0\npyparsing==3.0.9\nPyYAML==6.0\nCython\n" }, { "path": "scripts/__init__.py", "content": "" }, { "path": "scripts/config.yaml", "content": "paths:\n hero_cards_suits: '../images_templates/hero_cards_suits/'\n hero_cards_numbers: '../images_templates/hero_cards_numbers/'\n table_cards_numbers: '../images_templates/table_cards_numbers/'\n table_cards_suits: '../images_templates/table_cards_suits/'\n pot_numbers: '../images_templates/pot_numbers/'\n pot_image: '../images_templates/pot/pot.png'\n dealer_button: '../images_templates/dealer_button/button.png'\n empty_seat: '../images_templates/empty_seat/empty_seat.png'\n sitting_out: '../images_templates/sitting_out/so.png'\n\n\ntable_size:\n width: 1090\n height: 900\n\ninfo_box_size:\n width: 470\n height: 500\n\nhero_step_define:\n x_0: 511\n y_0: 605\n x_1: 528\n y_1: 644\n lower_gray_color: [0, 5, 50]\n upper_gray_color: [179, 10, 255]\n min_white_pixels: 400\n\n\nhero_cards:\n x_0: 478\n y_0: 535\n x_1: 616\n y_1: 592\n # x coordinate where the card ends\n separator_1: 67\n separator_2: 137\n\ntable_cards:\n x_0: 364\n y_0: 319\n x_1: 730\n y_1: 414\n # x coordinate where the card ends\n separator_1: 72\n separator_2: 144\n separator_3: 218\n separator_4: 290\n separator_5: 363\n\npot:\n # in what area looking for pot\n x_0: 472\n y_0: 279\n x_1: 636\n y_1: 328\n # the maximum width and height over which\n # the sum of the bet can be placed\n width: 100\n height: 20\n pot_template_width: 33\n\nplayer_center_coordinates:\n # the first player is the hero and then clockwise\n # coordinates are stored in [x, y] format\n 1: [469, 598]\n 2: [208,463]\n 3: [237,278]\n 4: [620,200]\n 5: [857, 274]\n 6: [867, 486]\n\nplayers_coordinates:\n # coordinates are stored in [x_0, y_0, x_1, y_1] format\n 1: [437,592,655,661]\n 2: [21,450, 245,525]\n 3: [49,214, 275,285]\n 4: [437, 139, 659, 213]\n 5: [814, 214, 1040,288]\n 6: [850, 449, 1075, 524]\n\nplayers_bet:\n # coordinates are stored in [x_0, y_0, x_1, y_1] format\n 2: [279, 457, 379, 480]\n 3: [292, 296, 392, 318]\n 4: [600, 222, 700, 246]\n 5: [688, 270, 788, 292]\n 6: [717, 457, 817, 480]" }, { "path": "scripts/equity.py", "content": "import numpy as np\nimport eval7\n\n\ndef calc_equity(deck, hero_cards, table_cards, iters=100000):\n \"\"\"\n Parameters:\n deck(list of str): all other cards, not including cards that are on the table and hero cards\n hero_cards(list of str): cards that belong to the hero\n table_cards(list of str): cards that are on the table\n iters(int): the amount that the table generates\n Returns:\n\n \"\"\"\n deck = [eval7.Card(card) for card in deck]\n table_cards = [eval7.Card(card) for card in table_cards]\n hero_cards = [eval7.Card(card) for card in hero_cards]\n max_table_cards = 5\n win_count = 0\n for _ in range(iters):\n np.random.shuffle(deck)\n num_remaining = max_table_cards - len(table_cards)\n draw = deck[:num_remaining+2]\n opp_hole, remaining_comm = draw[:2], draw[2:]\n player_hand = hero_cards + table_cards + remaining_comm\n opp_hand = opp_hole + table_cards + remaining_comm\n player_strength = eval7.evaluate(player_hand)\n opp_strength = eval7.evaluate(opp_hand)\n\n if player_strength > opp_strength:\n win_count += 1\n\n win_prob = (win_count / iters) * 100\n return round(win_prob, 2)\n\n" }, { "path": "scripts/grab_table.py", "content": "from PIL import Image\nimport numpy as np\nimport cv2\nfrom mss import mss\nfrom pokerstars_recognition import PokerStarsTableRecognizer\nfrom utils import read_config_file, set_window_size, remove_cards, data_concatenate\nfrom equity import calc_equity\nfrom info_box import update_label\n\nsct = mss()\nconfig = read_config_file()\nset_window_size()\n\ntable_data = []\nwhile True:\n updated_table_data = []\n monitor = {'top': 80, 'left': 70, 'width': config['table_size']['width'],\n 'height': config['table_size']['height']}\n img = Image.frombytes('RGB', (config['table_size']['width'], config['table_size']['height']),\n sct.grab(monitor).rgb)\n img = cv2.cvtColor(np.array(img), cv2.COLOR_BGR2RGB)\n recognizer = PokerStarsTableRecognizer(img, config)\n hero_step = recognizer.detect_hero_step()\n if hero_step:\n hero_cards = recognizer.detect_hero_cards()\n table_cards = recognizer.detect_table_cards()\n total_pot = recognizer.find_total_pot()\n updated_table_data.append([hero_cards, table_cards, total_pot])\n if table_data == updated_table_data:\n pass\n else:\n table_data = updated_table_data\n deck = remove_cards(hero_cards, table_cards)\n equity = calc_equity(deck, hero_cards, table_cards)\n players_info = recognizer.get_dealer_button_position()\n players_info = recognizer.get_empty_seats(players_info)\n players_info = recognizer.get_so_players(players_info)\n players_info = recognizer.assign_positions(players_info)\n players_info = recognizer.find_players_bet(players_info)\n text = data_concatenate(hero_cards, table_cards, total_pot, equity, players_info)\n update_label(text)\n\n" }, { "path": "scripts/info_box.py", "content": "from tkinter import *\nfrom scripts.utils import read_config_file\n\ncfg = read_config_file()\nroot = Tk()\nroot.geometry('{0}x{1}'.format(cfg['info_box_size']['width'], cfg['info_box_size']['height']))\nroot.title('PokerStarsHelper')\nroot.configure(background='ivory3')\nlab = Label(root, anchor=\"w\", justify=LEFT, font=(\"Arial\", 18))\nlab.pack(fill=\"both\", expand=True)\n\n\ndef update_label(text):\n lab.configure(text=text)\n root.update()" }, { "path": "scripts/pokerstars_recognition.py", "content": "import cv2\nfrom scripts.table_recognition import PokerTableRecognizer\nfrom scripts.utils import sort_bboxes, thresholding, card_separator, table_part_recognition, \\\n convert_contours_to_bboxes, find_by_template, find_closer_point, read_config_file\nimport numpy as np\n\n\nclass PokerStarsTableRecognizer(PokerTableRecognizer):\n\n def __init__(self, img, cfg):\n \"\"\"\n Parameters:\n img(numpy.ndarray): image of the whole table\n cfg (dict): config file\n \"\"\"\n self.img = img\n self.cfg = cfg\n\n def detect_hero_step(self):\n \"\"\"\n Based on the area under hero's cards,\n we determine whether hero should make a move\n Returns:\n Boolean Value(True or False): True, if hero step now\n \"\"\"\n res_img = self.img[self.cfg['hero_step_define']['y_0']:self.cfg['hero_step_define']['y_1'],\n self.cfg['hero_step_define']['x_0']:self.cfg['hero_step_define']['x_1']]\n\n hsv_img = cv2.cvtColor(res_img, cv2.COLOR_BGR2HSV_FULL)\n mask = cv2.inRange(hsv_img, np.array(self.cfg['hero_step_define']['lower_gray_color']),\n np.array(self.cfg['hero_step_define']['upper_gray_color']))\n count_of_white_pixels = cv2.countNonZero(mask)\n return True if count_of_white_pixels > self.cfg['hero_step_define']['min_white_pixels'] else False\n\n def detect_cards(self, separators, sort_bboxes_method, cards_coordinates, path_to_numbers, path_to_suits):\n \"\"\"\n Parameters:\n separators(list of int): contains values where the card ends\n sort_bboxes_method(str): defines how we will sort the contours.\n It can be left-to-right, bottom-to-top, top-to-bottom\n cards_coordinates(str): path to cards coordinates\n path_to_numbers(str): path where located numbers (J, K etc.)\n path_to_suits(str) : path where located suits\n Returns:\n cards_name(list of str): name of the cards\n \"\"\"\n cards_name = []\n img = self.img[self.cfg[cards_coordinates]['y_0']:self.cfg[cards_coordinates]['y_1'],\n self.cfg[cards_coordinates]['x_0']:self.cfg[cards_coordinates]['x_1']]\n binary_img = thresholding(img, 200, 255)\n contours, _ = cv2.findContours(binary_img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)\n bounding_boxes = convert_contours_to_bboxes(contours, 10, 2)\n bounding_boxes = sort_bboxes(bounding_boxes, method=sort_bboxes_method)\n cards_bboxes_dct = card_separator(bounding_boxes, separators)\n for _, cards_bboxes in cards_bboxes_dct.items():\n if len(cards_bboxes) == 3:\n cards_bboxes = [cards_bboxes[0]]\n\n elif len(cards_bboxes) == 0:\n return []\n\n elif len(cards_bboxes) > 3:\n raise ValueError(\"The number of bounding boxes should not be more than 3!\")\n\n card_name = ''\n for key, bbox in enumerate(cards_bboxes):\n color_of_img, directory = (cv2.IMREAD_COLOR, self.cfg['paths'][path_to_suits]) if key == 0 \\\n else (cv2.IMREAD_GRAYSCALE, self.cfg['paths'][path_to_numbers])\n res_img = img[bbox[1]:bbox[3], bbox[0]:bbox[2]]\n card_part = table_part_recognition(res_img, directory, color_of_img)\n card_name = card_part + 'T' if len(cards_bboxes) == 1 else card_name + card_part\n cards_name.append(card_name[::-1])\n return cards_name\n\n def detect_hero_cards(self):\n \"\"\"\n Returns:\n cards_name(list of str): name of the hero's cards\n \"\"\"\n separators = [self.cfg['hero_cards']['separator_1'], self.cfg['hero_cards']['separator_2']]\n sort_bboxes_method = 'bottom-to-top'\n cards_coordinates = 'hero_cards'\n path_to_numbers = 'hero_cards_numbers'\n path_to_suits = 'hero_cards_suits'\n cards_name = self.detect_cards(separators, sort_bboxes_method, cards_coordinates,\n path_to_numbers, path_to_suits)\n return cards_name\n\n def detect_table_cards(self):\n \"\"\"\n Returns:\n cards_name(list of str): name of the cards on the table\n \"\"\"\n separators = [self.cfg['table_cards']['separator_1'], self.cfg['table_cards']['separator_2'],\n self.cfg['table_cards']['separator_3'], self.cfg['table_cards']['separator_4'],\n self.cfg['table_cards']['separator_5']]\n sort_bboxes_method = 'top-to-bottom'\n cards_coordinates = 'table_cards'\n path_to_numbers = 'table_cards_numbers'\n path_to_suits = 'table_cards_suits'\n cards_name = self.detect_cards(separators, sort_bboxes_method, cards_coordinates,\n path_to_numbers, path_to_suits)\n return cards_name\n\n def find_total_pot(self):\n \"\"\"\n Returns:\n number(str): number with total pot\n \"\"\"\n img = self.img[self.cfg['pot']['y_0']:self.cfg['pot']['y_1'],\n self.cfg['pot']['x_0']:self.cfg['pot']['x_1']]\n _, max_loc = find_by_template(img, self.cfg['paths']['pot_image'])\n bet_img = img[max_loc[1] - 3:max_loc[1] + self.cfg['pot']['height'],\n max_loc[0] + self.cfg['pot']['pot_template_width']:\n max_loc[0] + self.cfg['pot']['pot_template_width'] + self.cfg['pot']['width']]\n binary_img = thresholding(bet_img, 105, 255)\n contours, _ = cv2.findContours(binary_img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)\n bounding_boxes = convert_contours_to_bboxes(contours, 3, 1)\n bounding_boxes = sort_bboxes(bounding_boxes, method='left-to-right')\n number = ''\n for bbox in bounding_boxes:\n number_img = bet_img[bbox[1]:bbox[3], bbox[0]:bbox[2]]\n symbol = table_part_recognition(number_img, self.cfg['paths']['pot_numbers'], cv2.IMREAD_GRAYSCALE)\n number += symbol\n return number\n\n def get_dealer_button_position(self):\n \"\"\"\n determine who is closer to the dealer button\n Returns:\n player_info(dict): here is information about all players as it becomes available\n \"\"\"\n player_info = {key: value for key in range(1, 7) for value in ['']}\n players_coordinates = self.cfg['player_center_coordinates']\n _, button_coordinates = find_by_template(self.img, self.cfg['paths']['dealer_button'])\n player_with_button = find_closer_point(players_coordinates, button_coordinates)\n player_info[player_with_button] = 'dealer_button'\n return player_info\n\n def get_missing_players(self, players_info, path_to_template_img, flag):\n \"\"\"\n find players who are currently absent for various reasons\n Parameters:\n players_info(dict): key - player number, value - '' - if the player is in the game;\n '-' - if a player's seat is available; '-so-' - if the player is absent\n path_to_template_img(str): the path to the images where the benchmark images are located\n flag(str): It can be - and -so-\n Returns:\n players_info(dict): info about players\n \"\"\"\n players_coordinates = self.cfg['players_coordinates']\n players_for_checking = [key for key, value in players_info.items() if value == '']\n for player, bbox in players_coordinates.items():\n if player != 1 and player in players_for_checking:\n player_img = self.img[bbox[1]:bbox[3], bbox[0]:bbox[2]]\n max_val, _ = find_by_template(player_img, self.cfg['paths'][path_to_template_img])\n if max_val > 0.8:\n players_info[player] = flag\n return players_info\n\n def get_empty_seats(self, players_info):\n \"\"\"\n find players whose places are currently vacant\n \"\"\"\n path_to_template_img = 'empty_seat'\n flag = '-'\n players_info = self.get_missing_players(players_info, path_to_template_img, flag)\n return players_info\n\n def get_so_players(self, players_info):\n \"\"\"\n find players who are not currently in the game\n \"\"\"\n path_to_template_img = 'sitting_out'\n flag = '-so-'\n players_info = self.get_missing_players(players_info, path_to_template_img, flag)\n return players_info\n\n def assign_positions(self, players_info):\n \"\"\"\n assign each player one of six positions if the player in the game\n Parameters:\n players_info(dict): info about players in {1:'', 2: 'dealer_button',3:'-so-' etc. } format\n Returns:\n players_info(dict): info about players in {1: 'BB', 2: 'SB', 3: '-so-' etc. } format\n \"\"\"\n busy_seats = [k for k, v in players_info.items() if v != '-' and v != '-so-']\n exist_positions = ['BTN', 'SB', 'BB', 'UTG', 'MP', 'CO']\n del exist_positions[3:3 + (6 - len(busy_seats))]\n player_with_button = [k for k, v in players_info.items() if v == 'dealer_button'][0]\n for index, player_number in enumerate(\n busy_seats[busy_seats.index(player_with_button):] + busy_seats[:busy_seats.index(player_with_button)]):\n if len(busy_seats) == 2:\n position = 'SB' if index == 0 else 'BB'\n players_info[player_number] = position\n else:\n players_info[player_number] = exist_positions[index]\n return players_info\n\n def find_players_bet(self, players_info):\n \"\"\"\n Parameters:\n players_info(dict): info about players in {1:'BTN', 2:'SB', 3:'BB' etc. } format\n Returns:\n updated_players_info(dict): info about players in {'Hero':'BTN', 'SB':'', 'BB':'50' etc. } format\n \"\"\"\n players_bet_location = self.cfg['players_bet']\n updated_players_info = {'Hero': players_info[1]}\n for i, location_coordinates in players_bet_location.items():\n if players_info[i] not in ('-so-', '-'):\n bet_img = self.img[location_coordinates[1]:location_coordinates[3],\n location_coordinates[0]:location_coordinates[2]]\n binary_img = thresholding(bet_img, 105, 255)\n contours, _ = cv2.findContours(binary_img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)\n bounding_boxes = convert_contours_to_bboxes(contours, 3, 1)\n bounding_boxes = sort_bboxes(bounding_boxes, method='left-to-right')\n number = ''\n for bbox in bounding_boxes:\n number_img = bet_img[bbox[1]:bbox[3], bbox[0]:bbox[2]]\n symbol = table_part_recognition(number_img, self.cfg['paths']['pot_numbers'], cv2.IMREAD_GRAYSCALE)\n number += symbol\n updated_players_info[players_info[i]] = number\n else:\n updated_players_info[i] = players_info[i]\n return updated_players_info\n\n\n" }, { "path": "scripts/table_recognition.py", "content": "from abc import ABC, abstractmethod\n\n\nclass PokerTableRecognizer(ABC):\n @abstractmethod\n def detect_hero_step(self):\n \"\"\"\n Based on the area under hero's cards,\n we determine whether hero should make a move\n \"\"\"\n pass\n\n @abstractmethod\n def detect_hero_cards(self):\n pass\n\n @abstractmethod\n def detect_table_cards(self):\n pass\n\n @abstractmethod\n def find_total_pot(self):\n pass\n\n @abstractmethod\n def get_dealer_button_position(self):\n \"\"\"\n determine who is closer to the dealer button\n \"\"\"\n pass\n\n @abstractmethod\n def get_empty_seats(self, players_info):\n \"\"\"\n find players whose places are currently vacant\n \"\"\"\n pass\n\n @abstractmethod\n def get_so_players(self, players_info):\n \"\"\"\n find players who are not currently in the game\n \"\"\"\n pass\n\n @abstractmethod\n def find_players_bet(self, players_info):\n pass\n\n\n\n\n\n\n\n" }, { "path": "scripts/utils.py", "content": "import cv2\nimport numpy as np\nimport yaml\nimport os\nfrom time import sleep\nfrom math import sqrt\n\n\ndef read_config_file(filename='config.yaml'):\n \"\"\"\n Parameters:\n filename (str): config file name\n Returns: loaded_data (dict)\n \"\"\"\n with open(filename, 'r') as stream:\n loaded_data = yaml.safe_load(stream)\n return loaded_data\n\n\ndef sort_bboxes(bounding_boxes, method):\n \"\"\"\n Parameters:\n bounding_boxes(list of lists of int): bounding_boxes in [x_0, y_0, x_1, y_1] format\n method(int): the method of sorting bounding boxes.\n It can be left-to-right, bottom-to-top or top-to-bottom\n Returns:\n bounding_boxes (list of tuple of int): sorted bounding boxes.\n Each bounding box presented in [x_0, y_0, x_1, y_1] format\n \"\"\"\n\n methods = ['left-to-right', 'bottom-to-top', 'top-to-bottom']\n if method not in methods:\n raise ValueError(\"Invalid method. Expected one of: %s\" % methods)\n\n else:\n\n if method == 'left-to-right':\n bounding_boxes.sort(key=lambda tup: tup[0])\n\n elif method == 'bottom-to-top':\n bounding_boxes.sort(key=lambda tup: tup[1], reverse=True)\n\n elif method == 'top-to-bottom':\n bounding_boxes.sort(key=lambda tup: tup[1], reverse=False)\n return bounding_boxes\n\n\ndef mse(img, benchmark_img):\n \"\"\"\n the 'Mean Squared Error' between two images that\n is the sum of the squared difference between the two images.\n NOTE: the two images must have the same dimension.\n Parameters:\n img(numpy.ndarray): image of a part of the table\n benchmark_img(numpy.ndarray): benchmark image.\n This image read from the folder.\n Returns:\n err (float): the error between two images\n \"\"\"\n err = np.sum((img.astype(\"float\") - benchmark_img.astype(\"float\")) ** 2)\n err /= float(img.shape[0] * img.shape[1])\n return err\n\n\ndef image_comparison(img, benchmark_img, color_of_img):\n \"\"\"\n Parameters:\n img(numpy.ndarray): image of a part of the table\n benchmark_img(str): path to benchmark image\n color_of_img(int): set in which format to read the image.\n It can be cv2.IMREAD_COLOR or cv2.IMREAD_GRAYSCALE\n Returns:\n err (float): the error between two images\n \"\"\"\n colors = [cv2.IMREAD_COLOR, cv2.IMREAD_GRAYSCALE]\n if color_of_img not in colors:\n raise ValueError(\"Invalid method. Expected one of: %s\" % colors)\n benchmark_img = cv2.imread(benchmark_img, color_of_img)\n res_img = cv2.resize(img, (benchmark_img.shape[1], benchmark_img.shape[0]))\n if color_of_img == cv2.IMREAD_GRAYSCALE:\n res_img = cv2.cvtColor(res_img, cv2.COLOR_BGR2GRAY)\n err = mse(res_img, benchmark_img)\n return err\n\n\ndef thresholding(img, value_1, value_2):\n \"\"\"\n Parameters:\n img(numpy.ndarray): image of a part of the table\n value_1(int): threshold value\n value_2(int): the maximum value that is assigned\n to pixel values that exceed the threshold value\n Returns: binary_img(numpy.ndarray)\n \"\"\"\n img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n _, binary_img = cv2.threshold(img, value_1, value_2, cv2.THRESH_BINARY)\n return binary_img\n\n\ndef table_part_recognition(img, directory, color_of_img):\n \"\"\"\n Parameters:\n img(numpy.ndarray): image of a part of the table\n directory(str): path to the template images\n color_of_img(int): set in which format to read the image.\n It can be cv2.IMREAD_COLOR or cv2.IMREAD_GRAYSCALE\n Returns:\n table_part(str): recognized suit, value (J, K etc.), pot etc.\n \"\"\"\n err_dict = {}\n for full_image_name in os.listdir(directory):\n image_name = full_image_name.split('.')[0]\n err = image_comparison(img, directory + full_image_name, color_of_img)\n err_dict[image_name] = err\n table_part = min(err_dict, key=err_dict.get)\n return table_part\n\n\ndef convert_contours_to_bboxes(contours, min_height, min_width):\n \"\"\"\n convert contours to bboxes, also remove all small bounding boxes\n Parameters:\n contours (tuple): each individual contour is a numpy array\n of (x, y) coordinates of boundary points of the object\n contours(list of tuples of int): bounding boxes suits\n and values (J, K etc.) in [x, y, w, h] format\n Returns:\n cards_bboxes(list of lists of int): bounding boxes suits\n and values (J, K etc.) in [x_0, y_0, x_1, y_1] format\n \"\"\"\n bboxes = [cv2.boundingRect(contour) for contour in contours]\n cards_bboxes = []\n for i in range(0, len(bboxes)):\n x, y, w, h = bboxes[i][0], bboxes[i][1], \\\n bboxes[i][2], bboxes[i][3]\n if h >= min_height and w >= min_width:\n contour_coordinates = [x - 1, y - 1, x + w + 1, y + h + 1]\n cards_bboxes.append(contour_coordinates)\n return cards_bboxes\n\n\ndef card_separator(bboxes, separators):\n \"\"\"\n determine which bounding box belongs to which card\n Parameters:\n bboxes(list of lists of int): bounding boxes suits and values (J, K etc.)\n separators(list of int): contains values where the card ends\n Returns:\n sorted_dct(dict) key - card number, value - bounding boxes\n \"\"\"\n dct = {}\n for bbox in bboxes:\n for separator in separators:\n if bbox[2] < separator:\n dct[separators.index(separator)] = dct.get(separators.index(separator), []) + [bbox]\n break\n sorted_dct = {key: value for key, value in sorted(dct.items(), key=lambda item: int(item[0]))}\n return sorted_dct\n\n\ndef find_by_template(img, path_to_image):\n \"\"\"\n Object detection using a \"template\".\n Parameters:\n img(numpy.ndarray): image of a part of the table/of the entire table\n path_to_image(str): the path to a template image\n Returns:\n max_val(float): largest value with the most likely match\n max_loc(tuple of int): location with the largest value.\n Location is presented in (x, y) format\n \"\"\"\n template_img_gray = cv2.imread(path_to_image, 0)\n img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n result = cv2.matchTemplate(img_gray, template_img_gray,\n cv2.TM_CCOEFF_NORMED)\n (min_val, max_val, min_loc, max_loc) = cv2.minMaxLoc(result)\n return max_val, max_loc\n\n\ndef load_images(directory):\n \"\"\"\n Parameters:\n directory(str): path to specific directory\n Returns:\n images(list of numpy.ndarray): images of a part of the table\n file_names(list of str): the name of the files in the folder\n \"\"\"\n images = [cv2.imread(directory + file) for file in sorted(os.listdir(directory))]\n file_names = [file for file in sorted(os.listdir(directory))]\n return images, file_names\n\n\ndef find_closer_point(players_coordinates, button_coordinates):\n \"\"\"\n find the distance between two points and find the minimum distance\n Parameters:\n players_coordinates(dict): key - player number, value - player coordinates\n button_coordinates(tuple): Coordinates are presented in (x, y) format\n Returns:\n player_with_button(int): the number of the player who is closest to the buttonthe number of the player who is closest to the button\n \"\"\"\n distance_dict = {}\n for player, player_coordinates in players_coordinates.items():\n distance = sqrt((player_coordinates[0] - button_coordinates[0]) ** 2\n + (player_coordinates[1] - button_coordinates[1]) ** 2)\n distance_dict[player] = distance\n player_with_button = min(distance_dict, key=distance_dict.get)\n return player_with_button\n\n\ndef remove_cards(hero_cards, table_cards):\n \"\"\"\n Parameters:\n hero_cards(list of str): cards that belong to the hero\n table_cards(list of str): cards that are on the table\n Returns:\n all_cards(list of str): all other cards, not including cards that are on the table and hero cards\n \"\"\"\n all_cards = ['2c', '2d', '2h', '2s', '3c', '3d', '3h', '3s', '4c', '4d', '4h', '4s', '5c', '5d', '5h', '5s',\n '6c', '6d', '6h', '6s', '7c', '7d', '7h', '7s', '8c', '8d', '8h', '8s', '9c', '9d', '9h', '9s',\n 'Tc', 'Td', 'Th', 'Ts', 'Jc', 'Jd', 'Jh', 'Js', 'Qc', 'Qd', 'Qh', 'Qs', 'Kc', 'Kd', 'Kh', 'Ks',\n 'Ac', 'Ad', 'Ah', 'As']\n for card in hero_cards+table_cards:\n all_cards.remove(card)\n return all_cards\n\n\ndef set_window_size():\n \"\"\"\n set the application window in the right place and with the right size\n \"\"\"\n sleep(3)\n cmd = 'wmctrl -r :ACTIVE: -e 0,0,0,1100,900'\n os.system(cmd)\n\n\ndef data_concatenate(hero_hand, table_cards, total_pot, equity, players_info):\n \"\"\"\n information from all lists, dictionaries are added to one common line\n \"\"\"\n text_players_info = ''\n for key, value in players_info.items():\n value = 'no bet found' if value == '' else value\n text_players_info += str(key) + ':' + str(value) + '\\n'\n\n table_cards = ['no cards on the table'] if table_cards == [] else table_cards\n\n text = 'Hero hand: ' + ' '.join(hero_hand) + '\\n' + 'Board: ' + ' '.join(table_cards) + '\\n' + \\\n 'Pot: ' + total_pot + '\\n' + 'Equity: ' + str(equity) + '%' + '\\n' + \\\n '------------------------------' + '\\n' + text_players_info\n return text\n\n\n" }, { "path": "unittests/TestTableRecognition.py", "content": "import unittest\nfrom scripts.pokerstars_recognition import PokerStarsTableRecognizer\nfrom scripts.utils import read_config_file, load_images\n\ncfg = read_config_file('../scripts/config.yaml')\ntest_cfg = read_config_file('test_config.yaml')\n\n\nclass TestTableRecognition(unittest.TestCase):\n\n def test_hero_step(self):\n images, file_names = load_images(test_cfg['paths']['hero_step'])\n for image, filename in zip(images, file_names):\n with self.subTest(\"TestHeroStep Incorrect detection in the image\", filename=filename):\n recognizer = PokerStarsTableRecognizer(image, cfg)\n self.assertEqual(recognizer.detect_hero_step(), test_cfg['hero_step'][filename])\n\n def test_hero_cards(self):\n images, file_names = load_images(test_cfg['paths']['hero_cards'])\n for image, filename in zip(images, file_names):\n with self.subTest(\"TestHeroCards Incorrect detection in the image\", filename=filename):\n recognizer = PokerStarsTableRecognizer(image, cfg)\n self.assertEqual(recognizer.detect_hero_cards(), test_cfg['hero_cards'][filename])\n\n def test_table_cards(self):\n images, file_names = load_images(test_cfg['paths']['table_cards'])\n for image, filename in zip(images, file_names):\n with self.subTest(\"TestTableCards Incorrect detection in the image\", filename=filename):\n recognizer = PokerStarsTableRecognizer(image, cfg)\n self.assertEqual(recognizer.detect_table_cards(), test_cfg['table_cards'][filename])\n\n def test_total_pot(self):\n images, file_names = load_images(test_cfg['paths']['total_pot'])\n for image, filename in zip(images, file_names):\n with self.subTest(\"TestTotalPot Incorrect detection in the image\", filename=filename):\n recognizer = PokerStarsTableRecognizer(image, cfg)\n self.assertEqual(recognizer.find_total_pot(), test_cfg['total_pot'][filename])\n\n def test_dealer_button_position(self):\n images, file_names = load_images(test_cfg['paths']['dealer_button_position'])\n for image, filename in zip(images, file_names):\n with self.subTest(\"TestDealerButton Incorrect detection in the image\", filename=filename):\n recognizer = PokerStarsTableRecognizer(image, cfg)\n self.assertEqual(recognizer.get_dealer_button_position(), test_cfg['dealer_button_position'][filename])\n\n def test_player_position(self):\n images, file_names = load_images(test_cfg['paths']['player_position'])\n for image, filename in zip(images, file_names):\n with self.subTest(\"TestPlayerPosition Incorrect detection in the image\", filename=filename):\n recognizer = PokerStarsTableRecognizer(image, cfg)\n players_info = recognizer.get_dealer_button_position()\n players_info = recognizer.get_empty_seats(players_info)\n players_info = recognizer.get_so_players(players_info)\n self.assertEqual(recognizer.assign_positions(players_info), test_cfg['player_position'][filename])\n\n def test_player_bet(self):\n images, file_names = load_images(test_cfg['paths']['player_bet'])\n for image, filename in zip(images, file_names):\n with self.subTest(\"TestPlayerBet Incorrect detection in the image\", filename=filename):\n recognizer = PokerStarsTableRecognizer(image, cfg)\n players_info = recognizer.get_dealer_button_position()\n players_info = recognizer.get_empty_seats(players_info)\n players_info = recognizer.get_so_players(players_info)\n players_info = recognizer.assign_positions(players_info)\n self.assertEqual(recognizer.find_players_bet(players_info), test_cfg['player_bet'][filename])\n\n\nif __name__ == '__main__':\n unittest.main()\n" }, { "path": "unittests/test_config.yaml", "content": "paths:\n hero_step: 'images_for_test/hero_step/'\n hero_cards: 'images_for_test/hero_cards/'\n table_cards: 'images_for_test/table_cards/'\n total_pot: 'images_for_test/total_pot/'\n dealer_button_position: 'images_for_test/dealer_button_position/'\n player_position: 'images_for_test/player_position/'\n player_bet: 'images_for_test/player_bet/'\n\nhero_step:\n test_image_1.png: True\n test_image_2.png: False\n test_image_3.png: True\n test_image_4.png: False\n test_image_5.png: False\n test_image_6.png: False\n test_image_7.png: False\n test_image_8.png: True\n test_image_9.png: True\n test_image_10.png: False\n test_image_11.png: False\n test_image_12.png: True\n test_image_13.png: True\n test_image_14.png: False\n test_image_15.png: False\n test_image_16.png: False\n test_image_17.png: True\n test_image_18.png: False\n test_image_19.png: True\n\nhero_cards:\n test_image_1.png: ['Qs', 'As']\n test_image_2.png: []\n test_image_3.png: ['7c','Kh']\n test_image_4.png: []\n test_image_5.png: ['2h','Ks']\n test_image_6.png: ['2h','Ks']\n test_image_7.png: ['Ah', '4h']\n test_image_8.png: ['Ah', '4h']\n test_image_9.png: ['2d', '6s']\n test_image_10.png: ['Qd', '3s']\n test_image_11.png: ['Kc', 'Ad']\n test_image_12.png: ['As', 'Td']\n test_image_13.png: ['5s', '9c']\n test_image_14.png: ['Kd', 'Kh']\n test_image_15.png: ['Ad', 'Td']\n test_image_16.png: ['2d', 'Tc']\n\ntable_cards:\n test_image_1.png: ['9h','5d','6d','9c']\n test_image_2.png: ['6c','4s','Ts','2s','8s']\n test_image_3.png: ['7h','4c','5h','Ac','Td']\n test_image_4.png: ['2d','3s','7h','Qh','Ah']\n test_image_5.png: ['Qc','Kd','2s']\n test_image_6.png: ['Th','Jc','Qc']\n test_image_7.png: ['Th','Jc','Qc','Jh']\n test_image_8.png: ['2s','Qh','Qs']\n test_image_9.png: ['2s','Qh','Qs', 'Ah']\n test_image_11.png: ['2s','Qh','Qs', 'Ah','Ac']\n test_image_12.png: ['Tc','9h','8d']\n test_image_13.png: ['5c','Qs','3d','2s']\n test_image_14.png: ['5c','Qs','3d','2s', 'Ah']\n test_image_15.png: ['Ad','Tc','4s','7h','3h']\n test_image_16.png: ['4c','7d','9d']\n test_image_17.png: ['9s','6c','7s','Jh']\n test_image_18.png: ['6d','Tc','5s']\n test_image_19.png: []\n test_image_20.png: ['3h','Qc','7c']\n\ntotal_pot:\n test_image_1.png: '1,550'\n test_image_2.png: '150'\n test_image_3.png: '2,268'\n test_image_4.png: '250'\n test_image_5.png: '861'\n test_image_7.png: '3,496'\n test_image_8.png: '17,450'\n test_image_9.png: '350'\n test_image_10.png: '2,354'\n test_image_11.png: '400'\n test_image_12.png: '750'\n test_image_13.png: '550'\n test_image_14.png: '38,169'\n test_image_15.png: '283'\n test_image_16.png: '478'\n\ndealer_button_position:\n test_image_1.png: {1: '', 2: 'dealer_button', 3: '', 4: '', 5: '', 6: ''}\n test_image_2.png: {1: '', 2: '', 3: 'dealer_button', 4: '', 5: '', 6: ''}\n test_image_3.png: {1: '', 2: '', 3: '', 4: 'dealer_button', 5: '', 6: ''}\n test_image_4.png: {1: '', 2: '', 3: '', 4: '', 5: '', 6: 'dealer_button'}\n test_image_5.png: {1: '', 2: '', 3: '', 4: 'dealer_button', 5: '', 6: ''}\n test_image_6.png: {1: 'dealer_button', 2: '', 3: '', 4: '', 5: '', 6: ''}\n test_image_7.png: {1: 'dealer_button', 2: '', 3: '', 4: '', 5: '', 6: ''}\n test_image_8.png: {1: '', 2: '', 3: '', 4: '', 5: 'dealer_button', 6: ''}\n test_image_9.png: {1: '', 2: '', 3: 'dealer_button', 4: '', 5: '', 6: ''}\n test_image_10.png: {1: '', 2: '', 3: '', 4: '', 5: '', 6: 'dealer_button'}\n test_image_11.png: {1: '', 2: '', 3: '', 4: 'dealer_button', 5: '', 6: ''}\n\nplayer_position:\n test_image_1.png: {1: 'CO', 2: 'BTN', 3: 'SB', 4: 'BB', 5: '-so-', 6: 'MP'}\n test_image_2.png: {1: 'MP', 2: 'CO', 3: 'BTN', 4: 'SB', 5: '-so-', 6: 'BB'}\n test_image_3.png: {1: 'BB', 2: '-so-', 3: 'CO', 4: 'BTN', 5: '-so-', 6: 'SB'}\n test_image_4.png: {1: 'SB', 2: 'BB', 3: 'MP', 4: 'CO', 5: '-so-', 6: 'BTN'}\n test_image_5.png: {1: 'BB', 2: 'CO', 3: '-so-', 4: '-so-', 5: 'BTN', 6: 'SB'}\n test_image_6.png: {1: 'BB', 2: 'MP', 3: 'CO', 4: 'BTN', 5: '-so-', 6: 'SB'}\n test_image_7.png: {1: 'BB', 2: '-so-', 3: '-so-', 4: '-', 5: '-', 6: 'SB'}\n test_image_8.png: {1: 'BB', 2: '-', 3: 'CO', 4: 'BTN', 5: '-', 6: 'SB'}\n test_image_9.png: {1: 'BTN', 2: 'SB', 3: 'BB', 4: 'UTG', 5: 'MP', 6: 'CO'}\n test_image_10.png: {1: 'BB', 2: '-', 3: 'CO', 4: 'BTN', 5: '-', 6: 'SB'}\n\nplayer_bet:\n test_image_1.png: {'Hero': 'CO', 'BTN': '', 'SB': '700', 'BB': '', 5: '-so-', 'MP': ''}\n test_image_2.png: {'Hero': 'SB', BB: '8,000', MP: '', CO: '', 5: '-so-', BTN: ''}\n test_image_3.png: {'Hero': 'BB', 2: '-so-', CO: '', BTN: '', 5: '-so-', SB: '50'}\n test_image_4.png: {'Hero': 'CO', BTN: '1,748', SB: '1,748', BB: '', 5: '-so-', MP: ''}\n test_image_5.png: {'Hero': 'SB', 2: '-', 'BB': '100', 'CO': '', 5: '-', BTN: '100'}\n test_image_6.png: {'Hero': 'BTN', 'SB': '50', 'BB': '100', 'MP': '100', 5: '-so-', CO: '100'}\n test_image_7.png: {'Hero': 'SB', 'BB': '', 3: '-so-', 4: '-so-', 'CO': '', 'BTN': ''}\n test_image_8.png: {'Hero': 'BB', 2: '-', 'CO': '', 'BTN': '100', 5: '-', 'SB': ''}\n test_image_9.png: {'Hero': 'BTN', 'SB': '', 'BB': '50', 'UTG': '100', 'MP': '200', 'CO': '200'}\n test_image_10.png: {'Hero': 'BB', 'CO': '2,362', 3: '-', 4: '-', 'BTN': '', 'SB': ''}\n\n\n\n\n\n" } ]