shriyaRam

from agora_community_sdk import AgoraRTC
client = AgoraRTC.create_watcher("app-id", "chromedriver link")
client.join_channel("channel-name")
users = client.get_users() # Gets references to everyone participating in the call
user1 = users[0] # reference the broadcasting user
binary_image = user1.frame # Gets the latest frame from the stream as a PIL image
binary_image.save("link to in.png file(high resolution)")

shriyaRam

def detect(im, param_vals):
    """
    Detect number plates in an image.
:param im:
        Image to detect number plates in.
:param param_vals:
        Model parameters to use. These are the parameters output by the `train`
        module.
:returns:
        Iterable of `bbox_tl, bbox_br, letter_probs`, defining the bounding box

shriyaRam

def letter_probs_to_code(letter_probs):
    return "".join(common.CHARS[i] for i in numpy.argmax(letter_probs, axis=1))

shriyaRam

if __name__ == "__main__":
    im = cv2.imread("link to in.png file")
    im_gray = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY) / 255.
f = numpy.load("link to weights.npz file")
    param_vals = [f[n] for n in sorted(f.files, key=lambda s: int(s[4:]))]
    print ("RUN")
for pt1, pt2, present_prob, letter_probs in post_process(
                                                  detect(im_gray, param_vals)):
        pt1 = tuple(reversed(map(int, pt1)))
        pt2 = tuple(reversed(map(int, pt2)))

shriyaRam

def _overlaps(match1, match2):
    bbox_tl1, bbox_br1, _, _ = match1
    bbox_tl2, bbox_br2, _, _ = match2
    return (bbox_br1[0] > bbox_tl2[0] and
            bbox_br2[0] > bbox_tl1[0] and
            bbox_br1[1] > bbox_tl2[1] and
            bbox_br2[1] > bbox_tl1[1])


def _group_overlapping_rectangles(matches):

shriyaRam

import pyttsx3
from predict import *
from agora_community_sdk import AgoraRTC
client = AgoraRTC.create_watcher("ab8078b5b58e4835af33a37580dbc864", "chromedriver.exe")
client.join_channel("test")
users = client.get_users() # Gets references to everyone participating in the call
user1 = users[0] # Can reference users in a list
binary_image = user1.frame
with open("ASL.jpg") as f:
    f.write(binary_image)

shriyaRam

# Initialise Text to speech engine
engine = pyttsx3.init()
engine.setProperty('rate', 105)
engine.setProperty('voice', 1)

shriyaRam

def withoutSkinSegment():
    window_name = "ASL"
    frame_height, frame_width, roi_height, roi_width = 480, 900, 200, 200
    cap = cv2.VideoCapture("ASL.jpg")
    cv2.namedWindow(window_name, cv2.WND_PROP_FULLSCREEN)
    x_start, y_start = 100, 100
    sentence = ""
    while True:
        ret, frame = cap.read()
        if ret is None:

shriyaRam

# Speak the sentence
        if len(sentence) > 0 and c == ord('s'):
            engine.say(sentence)
            engine.runAndWait()
        # Clear the sentence
        if c == ord('c') or c == ord('C'):
            sentence = ""
        # Delete the last character
        if c == ord('d') or c == ord('D'):
            sentence = sentence[:-1]

shriyaRam

"""
Contains functions : pre_process() and which() that are needed by translator.py for predicting image from webcam
"""
import cv2
import numpy as np
from variables import *
from keras.models import load_model


# Loads pretrained CNN Model from MODEL_PATH