File size: 3,106 Bytes
b641e9a 3956ae6 b641e9a e3fe776 b641e9a bca8baf b641e9a 3956ae6 87419c2 b641e9a 87419c2 b641e9a bca8baf b641e9a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 |
# Importing necessary libraries
import streamlit as st
import cv2
import json
import numpy as np
import threading
import time
import pandas as pd
import mediapipe as mp
import tensorflow as tf
import matplotlib.pyplot as plt
import streamlit.components.v1 as com
from mediapipe_functions import *
import tempfile
from streamlit_lottie import st_lottie
st.set_page_config(
page_title="Welcome to iSLR",
page_icon="👋",
)
# st.markdown("""
# <style>
# .css-9s5bis.edgvbvh3
# {
# visibility:hidden
# }
# .css-h5rgaw.egzxvld1
# {
# visibility:hidden
# }
# </style>
# """, unsafe_allow_html=True)
st.sidebar.success("Select a demo above.")
st.title("ASL Sign Language Recognition App")
f=st.file_uploader("Please upload a video of a demo of ASL sign", type=['mp4'])
if f is not None:
tfile = tempfile.NamedTemporaryFile(delete=False)
tfile.write(f.read())
width = max(30, 0.01)
side = max((100 - width) / 2, 0.01)
_, container, _ = st.columns([side, width, side])
container.video(data=tfile.name)
# st.video(tfile.name)
cap = cv2.VideoCapture(tfile.name)
stframe = st.empty()
# for local video file
# cap = cv2.VideoCapture('videos\goodbye.mp4')
final_landmarks=[]
with mp_holistic.Holistic(min_detection_confidence=0.5, min_tracking_confidence=0.5) as holistic:
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
image, results = mediapipe_detection(frame, holistic)
draw(image, results)
landmarks = extract_coordinates(results)
final_landmarks.extend(landmarks)
df1 = pd.DataFrame(final_landmarks,columns=['x','y','z'])
# Necessary functions
ROWS_PER_FRAME = 543
def load_relevant_data_subset(df):
data_columns = ['x', 'y', 'z']
data = df
n_frames = int(len(data) / ROWS_PER_FRAME)
data = data.values.reshape(n_frames, ROWS_PER_FRAME, len(data_columns))
return data.astype(np.float32)
# Loading data
test_df = load_relevant_data_subset(df1)
test_df = tf.convert_to_tensor(test_df)
# Inference
interpreter = tf.lite.Interpreter("models/model.tflite")
prediction_fn = interpreter.get_signature_runner("serving_default")
output = prediction_fn(inputs=test_df)
sign = np.argmax(output["outputs"])
sign_json=pd.read_json("sign_to_prediction_index_map.json",typ='series')
sign_df=pd.DataFrame(sign_json)
pred=sign_df.iloc[sign]
st.write(pred)
top_indices = np.argsort(output['outputs'])[::-1][:5]
top_values = output['outputs'][top_indices]
output_df = sign_df.iloc[top_indices]
output_df['Value'] = top_values
output_df.rename(columns = {0:'Index'}, inplace = True)
output_df.drop(['Index'],1, inplace=True)
st.write(output_df)
# callbacks : on_change, on_click
# com.iframe("https://embed.lottiefiles.com/animation/132349")
with open('assets/animations/14592-loader-cat.json') as source:
animation=json.load(source)
st_lottie(animation, width=300) |