Files added

README.md

@@ -1,3 +1,54 @@
-# Wall Crack Detection Model
-This model helps in detecting cracks or deforming walls.
-It is created by training CNNs on a large dataset of around 32,000 images (16000-Negative, 32000-Positive) Negative means it had no defect and Positive means it has defects.
+# Wall Crack Detection
+
+This project uses a deep learning model to detect cracks in walls using real-time video feed from a mobile phone camera. 
+The project is built using Tensorflow, Keras, and OpenCV.
+
+
+## Project Structure
+```
+|-- crack_detector.h5 # Trained Model
+|--main.py # Real-time detection script
+|--concrete_data/ #Dataset folder (if training from scratch is required)
+||--train/
+|||-Positive/
+|||-Negative/
+||--val/
+|||-Positive/
+|||-Negative/
+|--detection_log.csv # Optional, it will help you to store predictions
+|--README.md 
+```
+Requirements
+
+```
+- python 3.10+
+- Tensorflow 2.x
+- OpenCV
+- Numpy
+```
+Install Dependencies using pip:
+```
+pip install tensorflow opencv-python numpy
+```
+For better experience make a virtual environment.
+
+1. conda create --name <environment-name> python=3.10
+
+2. conda activate <environment-name> 
+conda deactivate
+
+conda activatae <environment-name>
+
+
+pip install tensorflow opencv-python numpy
+
+
+conda --version
+
+anaconda install 
+1. conda --version
+2. conda create --name <environment-name> python=3.10
+3. conda activate <environment-name>
+4. pip install tensorflow opencv-python numpy
+5. DroidCam IP in main.py file <url>
+6. run python main.py

main.py

@@ -0,0 +1,32 @@
+import cv2
+import tensorflow as tf
+import numpy as np
+model = tf.keras.models.load_model("crack_detector.h5")
+
+# Use HTTP instead of HTTPS
+url = "<Embedd-your-URL>" #Iinstall droidcam on phone and insert the URL
+cap = cv2.VideoCapture(url)
+
+while True:
+    ret, frame = cap.read()
+    if not ret:
+        print("Failed to grab frame. Check IP/Port and make sure phone/laptop are on same WiFi.")
+        break
+    
+    img = cv2.resize(frame, (224, 224))
+    img = img.astype("float32")/255.0
+    img = np.expand_dims(img, axis=0)
+
+    pred = model.predict(img, verbose=0)[0][0]
+    label = "No Crack" if pred<0.9 else "Crack Detected"
+
+    color = (0,255,0) if label=="No Crack" else (0,0,255)
+    cv2.putText(frame, label, (20,40), cv2.FONT_HERSHEY_SIMPLEX,1, color, 2)
+
+    cv2.imshow("Wall crack detection", frame)
+   
+    if cv2.waitKey(1) & 0xFF == ord('q'):
+        break
+
+cap.release()
+cv2.destroyAllWindows()

training.py

@@ -0,0 +1,48 @@
+import tensorflow as tf
+from tensorflow.keras.preprocessing.image import ImageDataGenerator
+from tensorflow.keras.applications import MobileNetV2
+from tensorflow.keras.layers import Dense, GlobalAveragePooling2D
+import os
+from tensorflow.keras.models import Model
+base_path = os.path.expanduser("~/Downloads/chirag-project/concrete_data")
+train_dir = os.path.join(base_path, "train")
+val_dir = os.path.join(base_path, "val")
+
+# Data generators
+datagen = ImageDataGenerator(rescale=1./255)
+
+train_gen = datagen.flow_from_directory(
+    train_dir,
+    target_size=(224, 224),
+    batch_size=32,
+    class_mode="binary"
+)
+
+val_gen = datagen.flow_from_directory(
+    val_dir,
+    target_size=(224, 224),
+    batch_size=32,
+    class_mode="binary"
+)
+
+# Base model
+base_model = MobileNetV2(weights="imagenet", include_top=False, input_shape=(224,224,3))
+x = base_model.output
+x = GlobalAveragePooling2D()(x)
+preds = Dense(1, activation="sigmoid")(x)
+
+model = Model(inputs=base_model.input, outputs=preds)
+
+# Freeze base layers for transfer learning
+for layer in base_model.layers:
+    layer.trainable = False
+
+model.compile(optimizer="adam", loss="binary_crossentropy", metrics=["accuracy"])
+
+# Train
+model.fit(train_gen, validation_data=val_gen, epochs=5)
+
+# Save model in repo
+model_save_path = os.path.expanduser("~/Downloads/crack_detector.h5")
+model.save(model_save_path)
+print(f"Model saved as {model_save_path}")