Merge remote-tracking branch 'origin/main'

Dockerfile

@@ -0,0 +1,17 @@
+# syntax=docker/dockerfile:1
+
+FROM python:3.11
+
+WORKDIR /code
+
+COPY requirements.txt .
+
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+COPY . .
+
+RUN chmod -R 755 combined_demo/input-imgs/ combined_demo/output-imgs/
+
+EXPOSE 3100
+
+CMD ["gunicorn", "main:app"]

color_analysis/color_analysis_model.py

@@ -135,7 +135,6 @@ def load_pretrained_model(model, save_path='color_analysis/trained_model.pth', d
     if os.path.exists(save_path):
         model.load_state_dict(torch.load(save_path, map_location=device, weights_only=True))
         model.eval()
-        print(f"Loaded pretrained model from {save_path}")
     else:
         print("No pretrained model found. Training from scratch.")
     return model
@@ -163,7 +162,6 @@ def main():
     ])
 
     # Create training and testing sets
-    print("Loading and splitting data...")
     train_set, test_set, train_labels, test_labels, class_labels = load_and_split_data(data_dir)
     train_dataset = ColorDataset(train_set, train_labels, transform=transform)
     test_dataset = ColorDataset(test_set, test_labels, transform=transform)
@@ -182,12 +180,10 @@ def main():
         print("Training the model...")
         train_model(model, train_loader, criterion, optimizer, num_epochs=num_epochs, device=device, save_path=save_path)
 
-    print("Testing the model...")
     test_loss, test_accuracy = test_model(model, test_loader, device=device)
     print(f"Test Loss: {test_loss:.4f}, Test Accuracy: {test_accuracy:.2f}%")
 
     # Predict the season of the input image with the trained model
-    print(f"Predicting season for image: {args.image_path}")
     predicted_season = predict_image(model, args.image_path, class_labels, transform, device=device)
     if predicted_season:
         print(f"Predicted Season: {predicted_season}")

face_detect/facedec/facedetect.py

@@ -3,9 +3,9 @@
 from typing import Tuple, Union
 
 import cv2
+import imutils
 import numpy as np
 from PIL import Image
-import imutils
 
 MARGIN = 10  # pixels
 ROW_SIZE = 10  # pixels
@@ -71,6 +71,8 @@ def visualize(image, detection_result) -> np.ndarray:
     return annotated_image, cropped
 
 
+import os
+
 # STEP 1: Import the necessary modules.
 import mediapipe as mp
 from mediapipe.tasks import python
@@ -96,14 +98,22 @@ def visualize(image, detection_result) -> np.ndarray:
 image_copy = np.copy(image.numpy_view())
 annotated_image, cropped = visualize(image_copy, detection_result)
 cropped = imutils.resize(cropped, width=800)
+cropped_pil = Image.fromarray(cv2.cvtColor(cropped, cv2.COLOR_BGR2RGB))
 img = cv2.imread(IMAGE_FILE)
 rgb_annotated_image = cv2.cvtColor(annotated_image, cv2.COLOR_BGR2RGB)
 rgb_annotated_image = imutils.resize(rgb_annotated_image, width=800)
+rgb_annotated_image_pil = Image.fromarray(cv2.cvtColor(rgb_annotated_image, cv2.COLOR_BGR2RGB))
 
 # SAVE CROPPED AS IMAGE: cv2.imwrite("cropped.jpg", cropped)
-
-cv2.imwrite("combined_demo/output-imgs/redbox.jpg", rgb_annotated_image)
-cv2.imwrite("combined_demo/output-imgs/cropped.jpg", cropped)
+# switch to os for saving images
+out_dir = "combined_demo/output-imgs"
+if not os.path.exists(out_dir):
+    os.mkdir(out_dir)
+rgb_annotated_image_pil.save(os.path.join(out_dir, "redbox.jpg"))
+cropped_pil.save(os.path.join(out_dir, "cropped.jpg"))
+
+# cv2.imwrite("combined_demo/output-imgs/redbox.jpg", rgb_annotated_image)
+# cv2.imwrite("combined_demo/output-imgs/cropped.jpg", cropped)
 
 # cv2.imshow("Gotchaface", rgb_annotated_image)
 # cv2.waitKey(0)

gunicorn.conf.py

@@ -0,0 +1,12 @@
+# Gunicorn configuration file
+import multiprocessing
+
+max_requests = 1000
+max_requests_jitter = 50
+
+log_file = "-"
+
+bind = "0.0.0.0:3100"
+
+worker_class = "uvicorn.workers.UvicornWorker"
+workers = (multiprocessing.cpu_count() * 2) + 1

requirements.txt

@@ -17,8 +17,4 @@ uvicorn[standard]
 sqlmodel
 gunicorn
 SQLAlchemy
-pydantic
-sklearn 
-argparse
-PIL
-os
+pydantic