Update git ignore, bring in missing files (calebolson123#20)

* update gitignore and add missing files

* add missing blanket_model dir

* empty the crop area

.dockerignore

@@ -0,0 +1,2 @@
+**/node_modules
+*.ipynb

.gitignore

@@ -15,3 +15,9 @@ build
 !.env_sample_docker
 !dockerfile
 !start_docker.sh
+!docker-compose.yml
+!notifications.txt
+!user_defined_crop_area.txt
+!.dockerignore
+!blanket_model
+!blanket_model/**

blanket_model/current_output/proba.txt

@@ -0,0 +1 @@
+0.45473471918237984,0.5452652808176202,1687742091.4165025

blanket_model/extract_inputs_from_video.py

@@ -0,0 +1,94 @@
+import cv2
+import os
+import sys
+import time
+from skimage.feature import hog
+from skimage import exposure
+from skimage import color
+import matplotlib.pyplot as plt
+from PIL import Image
+
+# Resizes a image and maintains aspect ratio
+def maintain_aspect_ratio_resize(image, width=None, height=None, inter=cv2.INTER_AREA):
+    # Grab the image size and initialize dimensions
+    dim = None
+    (h, w) = image.shape[:2]
+
+    # Return original image if no need to resize
+    if width is None and height is None:
+        return image
+
+    # We are resizing height if width is none
+    if width is None:
+        # Calculate the ratio of the height and construct the dimensions
+        r = height / float(h)
+        dim = (int(w * r), height)
+    # We are resizing width if height is none
+    else:
+        # Calculate the ratio of the 0idth and construct the dimensions
+        r = width / float(w)
+        dim = (width, int(h * r))
+
+    # Return the resized image
+    return cv2.resize(image, dim, interpolation=inter)
+
+input_path = sys.argv[1]
+output_path = sys.argv[2]
+
+def hog_it_up(image):
+    fd, hog_image = hog(color.rgb2gray(image), orientations=8, pixels_per_cell=(16, 16),
+                    cells_per_block=(1, 1), channel_axis=None, visualize=True)
+    hog_image_rescaled = exposure.rescale_intensity(hog_image, in_range=(0, 10))
+
+    # fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8, 4), sharex=True, sharey=True)
+    # ax1.axis('off')
+    # ax1.imshow(image, cmap=plt.cm.gray)
+    # ax1.set_title('Input image')
+    # # Rescale histogram for better display
+    # hog_image_rescaled = exposure.rescale_intensity(hog_image, in_range=(0, 10))
+    # ax2.axis('off')
+    # ax2.imshow(hog_image_rescaled, cmap=plt.cm.gray)
+    # ax2.set_title('Histogram of Oriented Gradients')
+    # plt.show()
+
+    return hog_image_rescaled
+
+x = 550
+y = 250
+h = 700
+w = 550
+
+frame_dim = (1920,1080)
+cap = cv2.VideoCapture(input_path)
+instance_count = 0
+fps = 60
+next_frame = 0
+count = 0
+counter = 1
+success = True
+while success:
+    success, frame = cap.read()
+    if count % fps == 0:
+        if frame.shape[0] > 1080 and frame.shape[1] > 1920:
+            frame = maintain_aspect_ratio_resize(frame.copy(), width=frame_dim[0], height=frame_dim[1])
+
+        cropped_frame = frame[y:y+h, x:x+w]
+
+    # TODO: test impacts of gaus blur
+        blur = cv2.GaussianBlur(cropped_frame,(5,5),0)
+
+        hoggy_cropped_frame = hog_it_up(blur)
+
+        file_name = input_path.rsplit('\\', 1)[-1] + '_frame' + str(instance_count) + '.jpg'
+
+        name = f'{output_path}\\{file_name}'
+        print ('Creating...' + name)
+
+        # cv2.imwrite(name, hoggy_cropped_frame)
+        plt.imsave(name, hoggy_cropped_frame)
+
+        instance_count += 1
+    count+=1
+
+cap.release()
+cv2.destroyAllWindows()

blanket_model/output/image_data.json

@@ -0,0 +1 @@
+{ "baby": [], "no_baby": [] }

blanket_model/train.py

@@ -0,0 +1,128 @@
+import os
+import cv2
+import argparse
+from sklearn.svm import LinearSVC
+from skimage import feature, exposure, color
+from skimage.feature import hog
+import matplotlib.pyplot as plt
+from PIL import Image
+
+# Resizes a image and maintains aspect ratio
+def maintain_aspect_ratio_resize(image, width=None, height=None, inter=cv2.INTER_AREA):
+    # Grab the image size and initialize dimensions
+    dim = None
+    (h, w) = image.shape[:2]
+
+    # Return original image if no need to resize
+    if width is None and height is None:
+        return image
+
+    # We are resizing height if width is none
+    if width is None:
+        # Calculate the ratio of the height and construct the dimensions
+        r = height / float(h)
+        dim = (int(w * r), height)
+    # We are resizing width if height is none
+    else:
+        # Calculate the ratio of the 0idth and construct the dimensions
+        r = width / float(w)
+        dim = (width, int(h * r))
+
+    # Return the resized image
+    return cv2.resize(image, dim, interpolation=inter)
+
+def hog_it_up(image):
+    fd, hog_image = hog(color.rgb2gray(image), orientations=8, pixels_per_cell=(16, 16),
+                    cells_per_block=(1, 1), channel_axis=None, visualize=True)
+    hog_image_rescaled = exposure.rescale_intensity(hog_image, in_range=(0, 10))
+
+    # fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8, 4), sharex=True, sharey=True)
+    # ax1.axis('off')
+    # ax1.imshow(image, cmap=plt.cm.gray)
+    # ax1.set_title('Input image')
+    # # Rescale histogram for better display
+    # hog_image_rescaled = exposure.rescale_intensity(hog_image, in_range=(0, 10))
+    # ax2.axis('off')
+    # ax2.imshow(hog_image_rescaled, cmap=plt.cm.gray)
+    # ax2.set_title('Histogram of Oriented Gradients')
+    # plt.show()
+
+    return fd, hog_image_rescaled
+
+x = 550
+y = 250
+h = 700
+w = 550
+frame_dim = (1920,1080)
+fps = 20
+def hog_the_video(input_path):
+    cap = cv2.VideoCapture(input_path)
+    instance_count = 0
+    next_frame = 0
+    count = 0
+    success = True
+    video_fds = []
+    video_images = []
+    while success:
+        success, frame = cap.read()
+        if count % fps == 0 and success:
+            if frame.shape[0] > 1080 and frame.shape[1] > 1920:
+                frame = maintain_aspect_ratio_resize(frame.copy(), width=frame_dim[0], height=frame_dim[1])
+
+            cropped_frame = frame[y:y+h, x:x+w]
+
+            fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8, 4), sharex=True, sharey=True)
+            ax1.axis('off')
+            ax1.imshow(cropped_frame, cmap=plt.cm.gray)
+            ax1.set_title('Input image')
+
+            cropped_frame_forced_size = cv2.resize(cropped_frame, (512, 512))
+            ax2.axis('off')
+            ax2.imshow(cropped_frame_forced_size, cmap=plt.cm.gray)
+            ax2.set_title('Histogram of Oriented Gradients')
+            plt.show()
+
+            # TODO: test impacts of gaus blur
+            blur = cv2.GaussianBlur(cropped_frame,(5,5),0)
+
+            fd, hoggy_cropped_frame = hog_it_up(blur)
+            print('shape: ', hoggy_cropped_frame.shape)
+
+            video_fds.append(fd)
+            video_images.append(hoggy_cropped_frame)
+
+            # file_name = input_path.rsplit('\\', 1)[-1] + '_frame' + str(instance_count) + '.jpg'
+            # name = f'{output_path}\\{file_name}'
+            # print ('Creating...' + name)
+            # cv2.imwrite(name, hoggy_cropped_frame)
+            # plt.imsave(name, hoggy_cropped_frame)
+            # instance_count += 1
+        count+=1
+
+    cap.release()
+    cv2.destroyAllWindows()
+    return video_fds, video_images
+
+### iterate over raw video inputs, select images throughout video, get the hog feature descriptors & images 
+all_images = []
+all_fds = []
+all_labels = []
+video_paths = os.listdir(f"./raw")
+for path in video_paths:
+    all_videos = os.listdir(f"raw/{path}")
+    for video in all_videos:
+        video_path = f"./raw/{path}/{video}"
+        print('video_path: ', video_path)
+
+        fds, images = hog_the_video(video_path)
+
+        # image = cv2.resize(image, (128, 256))
+        # get the HOG descriptor for the image
+        # hog_desc = feature.hog(image, orientations=9, pixels_per_cell=(8, 8),
+        #     cells_per_block=(2, 2), transform_sqrt=True, block_norm='L2-Hys')
+
+        print(path)
+        print(images)
+        all_images = all_images + images
+        all_fds = all_fds + fds
+        all_labels = all_labels + [path] * len(fds)

docker-compose.yml

@@ -0,0 +1,18 @@
+services:
+
+  baby-sleep-coach:
+    build:
+      dockerfile: ./Dockerfile
+      context: .
+    ports:
+      - "0.0.0.0:${PORT}:80"
+      - "0.0.0.0:8000:8000"
+      - "0.0.0.0:8001:8001"
+    env_file:
+      - .env
+    volumes:
+      - type: bind
+        # You need to update this to absolute path of this dir on your machine
+        source: C:\Users\caleb\projects\github-sleepy-baby\BabySleepCoach\
+        target: "${APP_DIR}"
+    container_name: baby-sleep-coach

notifications.txt

@@ -0,0 +1 @@
+false