Update pages/4_Writing.py

pages/4_Writing.py

@@ -3,6 +3,8 @@ from streamlit_drawable_canvas import st_canvas
 import cv2
 from tensorflow.keras.models import load_model
 import numpy as np
+import random
+import os
 
 
 arabic_chars = ['alef','beh','teh','theh','jeem','hah','khah','dal','thal','reh','zain','seen','sheen',
@@ -34,41 +36,89 @@ def add_logo():
  unsafe_allow_html=True,
  )
 add_logo()
+
+
+
+
 def predict_image(image_path, model_path):
- model = load_model(model_path)
-
- img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
- img = cv2.resize(img, (32, 32))
- img = img.reshape(1, 32, 32, 1)
- img = img.astype('float32') / 255.0
-
- pred = model.predict(img)
- predicted_label = arabic_chars[np.argmax(pred)]
-
- return predicted_label
-
-canvas_result = st_canvas(
- fill_color="rgba(255, 255, 255, 0.3)", # Filled color (white)
- stroke_width=30, # Stroke width
- stroke_color="#FFFFFF", # Stroke color (white)
- background_color="#000000", # Canvas background color (black)
- update_streamlit=True,
- height=400,
- width=400,
- drawing_mode="freedraw",
- key="canvas",
-)
-
-if st.button("Predict"):
- if canvas_result.image_data is not None:
- image = canvas_result.image_data
- image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
- image = cv2.resize(image, (32, 32))
- cv2.imwrite("temp_image.png", image)
-
- model_path = "saved_model.h5" # Replace with the path to your trained model
- predicted_label = predict_image("temp_image.png", model_path)
-
- st.write(f"Predicted Character: {predicted_label}")
+ try:
+ model = load_model(model_path)
+ except Exception as e:
+ st.error(f"Error loading model: {e}")
+ return None
+
+ try:
+ img = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
+ img = cv2.resize(img, (32, 32))
+ img = img.reshape(1, 32, 32, 1)
+ img = img.astype('float32') / 255.0
+
+ pred = model.predict(img)
+ predicted_label = arabic_chars[np.argmax(pred)]
+ return predicted_label
+ except Exception as e:
+ st.error(f"Error processing image: {e}")
+ return None
+
+def get_random_image(folder_path):
+ try:
+ char = random.choice(arabic_chars)
+ image_path = os.path.join(folder_path, f"{char}.png")
+ return image_path, char
+ except Exception as e:
+ st.error(f"Error loading random image: {e}")
+ return None, None
+
+# Streamlit app
+st.title("Arabic Character Recognition")
+
+# Load and display a random image
+folder_path = "arabic letters"
+if 'image_path' not in st.session_state:
+ st.session_state.image_path, st.session_state.correct_char = get_random_image(folder_path)
+col1,col2,col3=st.columns([1,1,1])
+with col1:
+ if st.session_state.image_path and st.session_state.correct_char:
+ st.image(st.session_state.image_path, caption=f"Draw this character: {st.session_state.correct_char}",width=350,)
  else:
- st.write("Please draw something on the canvas.")
+ st.error("Error loading the random image.")
+
+with col2:
+ canvas_result = st_canvas(
+ fill_color="rgba(255, 255, 255, 0.3)", # Filled color (white)
+ stroke_width=19, # Stroke width
+ stroke_color="#FFFFFF", # Stroke color (white)
+ background_color="#000000", # Canvas background color (black)
+ update_streamlit=True,
+ height=400,
+ width=400,
+ drawing_mode="freedraw",
+ key="canvas",
+ )
+with col3:
+ if st.button("Check"):
+ if canvas_result.image_data is not None:
+ image = canvas_result.image_data
+ image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
+ image = cv2.resize(image, (32, 32))
+
+ # Save the image temporarily
+ temp_image_path = "temp_image.png"
+ cv2.imwrite(temp_image_path, image)
+
+ # Load and predict using the model
+ model_path = "saved_model.h5" # Replace with the path to your trained model
+ if os.path.exists(model_path):
+ predicted_label = predict_image(temp_image_path, model_path)
+ if predicted_label:
+ #st.write(f"Predicted Character: {predicted_label}")
+ if predicted_label == st.session_state.correct_char:
+ st.success("You are correct!")
+ st.session_state.image_path, st.session_state.correct_char = get_random_image(folder_path)
+ canvas_result.clear_background()
+ else:
+ st.error("The prediction does not match the displayed character. Try again.")
+ else:
+ st.error("Model file not found. Please check the model path.")
+ else:
+ st.write("Please draw something on the canvas.")