these are the model details Model: "sequential_1" ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩ │ dense_3 (Dense) │ (None, 128) │ 100,480 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ batch_normalization │ (None, 128) │ 512 │ │ (BatchNormalization) │ │ │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dropout (Dropout) │ (None, 128) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dense_4 (Dense) │ (None, 64) │ 8,256 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ batch_normalization_1 │ (None, 64) │ 256 │ │ (BatchNormalization) │ │ │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dropout_1 (Dropout) │ (None, 64) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dense_5 (Dense) │ (None, 10) │ 650 │ └──────────────────────────────────────┴─────────────────────────────┴─────────────────┘ Total params: 110,154 (430.29 KB) Trainable params: 109,770 (428.79 KB) Non-trainable params: 384 (1.50 KB) inal training accuracy: 0.9999 Final validation accuracy: 1.0000 79/79 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step Confusion Matrix: |[[254 0 0 0 0 0 0 0 0 0]| |[ 0 253 0 0 0 0 0 0 0 0]| |[ 0 0 252 0 0 0 0 0 0 0]| |[ 0 0 0 253 0 0 0 0 0 0]| |[ 0 0 0 0 253 0 0 0 0 0]| |[ 0 0 0 0 0 253 0 0 0 0]| |[ 0 0 0 0 0 0 0 253 0 0]| |[ 0 0 0 0 0 0 0 0 252 0]| |[ 0 0 0 0 0 0 0 0 0 253]]| Classification Report: precision recall f1-score support 0 1.00 1.00 1.00 254 1 1.00 1.00 1.00 253 2 1.00 1.00 1.00 252 3 1.00 1.00 1.00 253 4 1.00 1.00 1.00 253 5 1.00 1.00 1.00 253 6 1.00 1.00 1.00 252 7 1.00 1.00 1.00 253 8 1.00 1.00 1.00 252 9 1.00 1.00 1.00 253 accuracy 1.00 2528 macro avg 1.00 1.00 1.00 2528 weighted avg 1.00 1.00 1.00 2528 ``` # Convert the BGR image to RGB rgb_img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) # Process the image and find hands results = hands.process(rgb_img) gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # Check if hand landmarks are detected if results.multi_hand_landmarks: for hand_landmarks in results.multi_hand_landmarks: # Draw hand landmarks mp_drawing.draw_landmarks(img, hand_landmarks, mp_hands.HAND_CONNECTIONS) # Calculate bounding box around the hand h, w, c = img.shape x_max = 0 y_max = 0 x_min = w y_min = h for lm in hand_landmarks.landmark: x, y = int(lm.x * w), int(lm.y * h) if x > x_max: x_max = x if x < x_min: x_min = x if y > y_max: y_max = y if y < y_min: y_min = y # Add padding to the bounding box padding = 20 x_min = max(0, x_min - padding) y_min = max(0, y_min - padding) x_max = min(w, x_max + padding) y_max = min(h, y_max + padding) # Draw rectangle around hand cv2.rectangle(img, (x_min, y_min), (x_max, y_max), color_dict, 2) # Crop the hand region if x_min < x_max and y_min < y_max: # Ensure valid crop region crop_img = gray[y_min:y_max, x_min:x_max] if crop_img.size != 0: # Check if crop is not empty count = count + 1 if(count % 100 == 0): prev_val = count # Process the cropped hand image blur = cv2.GaussianBlur(crop_img, (5, 5), 2) th3 = cv2.adaptiveThreshold(blur, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY_INV, 11, 2) ret, res = cv2.threshold(th3, minValue, 255, cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU) # Resize to match model input size resized = cv2.resize(res, (img_size, img_size)) normalized = resized/255.0 reshaped = np.reshape(normalized, (1, img_size, img_size, 1)) # Predict with model result = model.predict(reshaped) label = np.argmax(result, axis=1)[0] if(count == 300): count = 99 prev = labels_dict[label] if(label == 0): string = string + " " else: string = string + prev # Show the processed hand image cv2.imshow("Hand", res) else: # If no hand is detected, reset counter count = 0 cv2.putText(img, prev, (24, 14), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (255, 255, 255), 2) cv2.putText(img, string, (275, 50), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (200, 200, 200), 2) #cv2.putText(img, str(prev_val//100), (300, 150), cv2.FONT_HERSHEY_SIMPLEX, 1.5, (255, 255, 255), 2) cv2.imshow('LIVE', img) key = cv2.waitKey(1) if(key == 27): # press Esc. to exit break ``` ` whenever the cell is run it accurately recognizes the hand and moves around with it but is does not convert the gestures to alphabets Continue reading...
