CircumSpect / dense.py
crystal-technologies's picture
Update dense.py
6e2e3a9
raw
history blame
2.45 kB
from matplotlib.patches import Rectangle
from describe import process_image
import matplotlib.pyplot as plt
from PIL import Image
import ocrmac
import json
import time
import cv2
import os
time.sleep(2)
with open('pwd.txt', 'r') as pwd:
folder_location = pwd.read()
def crop_and_save_image(img, box, output_path):
# Convert box coordinates to integers
box = [int(coord) for coord in box]
# Crop the image to the specified region of interest
cropped_img = img[box[1]:box[3], box[0]:box[2]]
cv2.imwrite(output_path, cropped_img)
def visualize_result(image_file_path, result):
assert isinstance(result, list)
og_img = cv2.imread(image_file_path)
img = cv2.imread(image_file_path)
captions = []
for r in result:
box = r['box']
caption = r['cap']
if "<unk>" in caption:
crop_and_save_image(og_img, box, "ocr.png")
recognized = ocrmac.OCR('Sample Images/Image.jpeg').recognize()
caption = caption.replace("<unk>", recognized[0][0])
cv2.rectangle(img, (int(box[0]), int(box[1])),
(int(box[2]), int(box[3])), (0, 0, 255), 2)
cv2.rectangle(img, (int(box[0]), int(box[1])), (int(
box[2]), int(box[1])-50), (200, 200, 200), -1)
cv2.rectangle(img, (int(box[0]), int(box[1])),
(int(box[2]), int(box[1])-50), (0, 0, 0), 2)
cv2.putText(img, caption, (int(box[0]), int(
box[1]) - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0), 2)
captions.append(caption)
cv2.imwrite("output.png", img)
return captions, img
def describe_image(frame):
IMG_FILE_PATH = f'{folder_location}image.png'
cv2.imwrite(IMG_FILE_PATH, frame)
process_image(IMG_FILE_PATH, folder_location)
RESULT_JSON_PATH = f'{folder_location}CircumSpect/result.json'
with open(RESULT_JSON_PATH, 'r') as f:
results = json.load(f)
TO_K = 10
assert IMG_FILE_PATH in results.keys()
captions, frame = visualize_result(
IMG_FILE_PATH, results[IMG_FILE_PATH][:TO_K])
return captions, frame
if __name__ == "__main__":
cap = cv2.VideoCapture(0)
time.sleep(2)
start = time.time()
while True:
end = time.time()
print(end-start)
_, img = cap.read()
caption, frame = describe_image(img)
cv2.imshow("CircumSpect", frame)
cv2.waitKey(1)
start = time.time()