slashtechno
/
wyzely-detect
mirror of https://github.com/slashtechno/wyzely-detect
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Simple object detection

This commit is contained in:
slashtechno 2023-10-01 19:56:40 -05:00
parent 3a2ed7d4eb
commit c56d7c86fc
Signed by: slashtechno
GPG Key ID: 8EC1D9D9286C2B17
12 changed files with 2507 additions and 221 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
.gitignore vendored
View File

@ -1,2 +1,3 @@
.env
config/
using_yolov8.ipynb

1
.python-version Normal file
View File

@ -0,0 +1 @@
3.10.5

BIN
environment.yml
View File

Binary file not shown.

209
main.py
View File

@ -1,209 +0,0 @@
import datetime
import face_recognition
import cv2
import numpy as np
from dotenv import load_dotenv
import os
import json
import pathlib
import requests
import time
load_dotenv()
URL = os.getenv("URL")
RUN_SCALE = os.getenv("RUN_SCALE")
VIEW_SCALE = os.getenv("VIEW_SCALE")
DATETIME_FORMAT = "%Y-%m-%d %H:%M:%S"
# RUN_SCALE = 0.25
# VIEW_SCALE = 0.75
DISPLAY = False
RUN_BY_COMPOSE = os.getenv("RUN_BY_COMPOSE")
NTFY_URL = os.getenv("NTFY_URL")
def find_face_from_name(name):
for face in config["faces"]:
if config["faces"][face]["name"] == name:
return face
return None
def write_config():
with open(config_path, "w") as config_file:
json.dump(config, config_file, indent=4)
print("Hello, world!")
# Initialize some variables
face_locations = []
face_encodings = []
face_names = []
known_face_encodings = []
known_face_names = []
process_this_frame = True
# Load the config file, if it does not exist or is blank, create it
config = {
# If RUN_BY_COMPOSE is true, set url to rtsp://wyze-bridge:8554/wyze_cam_name, otherwise set it to "rtsp://localhost:8554/wyze_cam_name"
"URL": "rtsp://localhost:8554/wyze_cam_name"
if not RUN_BY_COMPOSE
else "rtsp://bridge:8554/wyze_cam_name",
"run_scale": "0.25",
"view_scale": "0.75",
"faces": {
"example1": {"image": "config/example1.jpg", "last_seen": ""},
"example2": {"image": "config/example2.jpg", "last_seen": ""},
},
"ntfy_url": "https://ntfy.sh/example",
"display": True,
}
config_path = pathlib.Path("config/config.json")
if config_path.exists():
with open(config_path, "r") as config_file:
config = json.load(config_file)
else:
with open(config_path, "w") as config_file:
json.dump(config, config_file, indent=4)
print("Config file created, please edit it and restart the program")
print("For relative paths, use the format config/example.jpg")
exit()
if URL:
config["URL"] = URL
else:
URL = config["URL"]
if RUN_SCALE:
config["RUN_SCALE"] = RUN_SCALE
else:
RUN_SCALE = float(config["RUN_SCALE"])
if VIEW_SCALE:
config["VIEW_SCALE"] = VIEW_SCALE
else:
VIEW_SCALE = float(config["VIEW_SCALE"])
if DISPLAY:
config["DISPLAY"] = DISPLAY
else:
DISPLAY = config["display"]
if NTFY_URL:
config["ntfy_url"] = NTFY_URL
else:
NTFY_URL = config["ntfy_url"]
print(f"Current config: {config}")
for face in config["faces"]:
# Load a sample picture and learn how to recognize it.
image = face_recognition.load_image_file(config["faces"][face]["image"])
face_encoding = face_recognition.face_encodings(image)[0]
known_face_encodings.append(face_encoding)
# Append the key to the list of known face names
known_face_names.append(face)
video_capture = cv2.VideoCapture(URL)
# Eliminate lag by setting the buffer size to 1
# This makes it so that the video capture will only grab the most recent frame
# However, this means that the video may be choppy
video_capture.set(cv2.CAP_PROP_BUFFERSIZE, 1)
# Print the resolution of the video
print(
f"Video resolution: {video_capture.get(cv2.CAP_PROP_FRAME_WIDTH)}x{video_capture.get(cv2.CAP_PROP_FRAME_HEIGHT)}"
)
print("Beginning video capture...")
while True:
# Grab a single frame of video
ret, frame = video_capture.read()
# Only process every other frame of video to save time
# Resize frame of video to a smaller size for faster face recognition processing
run_frame = cv2.resize(frame, (0, 0), fx=RUN_SCALE, fy=RUN_SCALE)
view_frame = cv2.resize(frame, (0, 0), fx=VIEW_SCALE, fy=VIEW_SCALE)
# Convert the image from BGR color (which OpenCV uses) to RGB color (which face_recognition uses)
rgb_run_frame = run_frame[:, :, ::-1]
# Find all the faces and face encodings in the current frame of video
# model cnn is gpu accelerated, but hog is cpu only
face_locations = face_recognition.face_locations(
rgb_run_frame, model="hog"
) # This crashes the program without output on my laptop when it's running without Docker compose
face_encodings = face_recognition.face_encodings(rgb_run_frame, face_locations)
face_names = []
for face_encoding in face_encodings:
# See if the face is a match for the known face(s)
matches = face_recognition.compare_faces(known_face_encodings, face_encoding)
name = "Unknown"
# Or instead, use the known face with the smallest distance to the new face
face_distances = face_recognition.face_distance(
known_face_encodings, face_encoding
)
best_match_index = np.argmin(face_distances)
if matches[best_match_index]:
name = known_face_names[best_match_index]
last_seen = config["faces"][name]["last_seen"]
# If it's never been seen, set the last seen time to x+5 seconds ago so it will be seen
# Kind of a hacky way to do it, but it works... hopefully
if last_seen == "":
print(f"{name} has been seen for the first time")
config["faces"][name]["last_seen"] = (
datetime.datetime.now() - datetime.timedelta(seconds=15)
).strftime(DATETIME_FORMAT)
write_config()
# Check if the face has been seen in the last 5 seconds
if datetime.datetime.now() - datetime.datetime.strptime(
last_seen, DATETIME_FORMAT
) > datetime.timedelta(seconds=10):
print(f"{name} has been seen")
# Send a notification
print(f"Sending notification to{NTFY_URL}")
requests.post(
NTFY_URL,
data=f'"{name}" has been seen',
headers={
"Title": "Face Detected",
"Priority": "default",
"Tags": "neutral_face",
},
)
# Update the last seen time
config["faces"][name]["last_seen"] = datetime.datetime.now().strftime(
DATETIME_FORMAT
)
# print("Writing config...")
write_config()
face_names.append(name)
# Display the results
# Iterate over each face found in the frame to draw a box around it
# Zip is used to iterate over two lists at the same time
for (top, right, bottom, left), name in zip(face_locations, face_names):
# print(f"Face found at {top}, {right}, {bottom}, {left} with name {name}")
# Scale back up face locations since the frame we detected in was scaled to 1/4 size
top = int(top * (VIEW_SCALE / RUN_SCALE))
right = int(right * (VIEW_SCALE / RUN_SCALE))
bottom = int(bottom * (VIEW_SCALE / RUN_SCALE))
left = int(left * (VIEW_SCALE / RUN_SCALE))
# Draw a box around the face
cv2.rectangle(view_frame, (left, top), (right, bottom), (0, 0, 255), 2)
# Draw a label with a name below the face
cv2.rectangle(
view_frame, (left, bottom - 35), (right, bottom), (0, 0, 255), cv2.FILLED
)
font = cv2.FONT_HERSHEY_DUPLEX
cv2.putText(
view_frame, name, (left + 6, bottom - 6), font, 1.0, (255, 255, 255), 1
)
# Display the resulting image if DISPLAY is set to true
if config["display"]:
cv2.imshow("Scaled View", view_frame)
# Hit 'q' on the keyboard to quit!
if cv2.waitKey(1) & 0xFF == ord("q"):
break
# Release handle to the webcam
print("Releasing video capture")
video_capture.release()
cv2.destroyAllWindows()

2275
poetry.lock generated Normal file
View File

File diff suppressed because it is too large Load Diff

58
pyproject.toml Normal file
View File

@ -0,0 +1,58 @@
[tool.poetry]
name = "detect-it"
version = "0.1.0"
description = "Detect all the things"
authors = ["slashtechno <77907286+slashtechno@users.noreply.github.com>"]
license = "MIT"
readme = "README.md"
packages = [{include = "detect_it"}]
[tool.poetry.dependencies]
python = "^3.10"
python-dotenv = "^1.0.0"
httpx = "^0.25.0"
opencv-python = "^4.8.1.78"
ultralytics = "^8.0.190"
hjson = "^3.1.0"
numpy = "^1.23.2"
torch = [
{ version = "^2.0.0+cu118", source = "torch_cu118", markers = "extra=='cuda'" },
{ version = "^2.0.0+cpu", source = "torch_cpu", markers = "extra!='cuda'" },
]
torchaudio = [
{ version = "^2.0.0+cu118", source = "torch_cu118", markers = "extra=='cuda'" },
{ version = "^2.0.0+cpu", source = "torch_cpu", markers = "extra!='cuda'" },
]
torchvision = [
{ version = "^0.15+cu118", source = "torch_cu118", markers = "extra=='cuda'" },
{ version = "^0.15+cpu", source = "torch_cpu", markers = "extra!='cuda'" },
]
[tool.poetry.group.dev.dependencies]
black = "^23.9.1"
ruff = "^0.0.291"
ipykernel = "^6.25.2"
[[tool.poetry.source]]
name = "torch_cpu"
url = "https://download.pytorch.org/whl/cpu"
priority = "supplemental"
[[tool.poetry.source]]
name = "torch_cu118"
url = "https://download.pytorch.org/whl/cu118"
priority = "supplemental"
[tool.poetry.extras]
cuda = []
[[tool.poetry.source]]
name = "PyPI"
priority = "primary"
[build-system]
requires = ["poetry-core"]
build-backend = "poetry.core.masonry.api"

11
requirements.txt
View File

@ -1,11 +0,0 @@
# certifi @ file:///croot/certifi_1665076670883/work/certifi
click==8.1.3
dlib==19.24.0
face-recognition==1.3.0
face-recognition-models==0.3.0
numpy==1.23.5
opencv-python==4.6.0.66
Pillow==9.3.0
python-dotenv==0.21.0
urllib3==1.26.13
requests==2.31.0

0
src/__init__.py Normal file
View File

129
src/__main__.py Normal file
View File

@ -0,0 +1,129 @@
# import face_recognition
import cv2
import numpy as np
import dotenv
from pathlib import Path
import os
import time
# import hjson as json
import torch
from ultralytics import YOLO
import argparse
from .utils import notify, config_utils
DATETIME_FORMAT = "%Y-%m-%d %H:%M:%S"
args = None
def main():
global args
# RUN_BY_COMPOSE = os.getenv("RUN_BY_COMPOSE") # Replace this with code to check for gpu
if Path(".env").is_file():
dotenv.load_dotenv()
print("Loaded .env file")
else:
print("No .env file found")
argparser = argparse.ArgumentParser(
prog="Detect It",
description="Detect it all!",
epilog=":)",
)
# required='RUN_SCALE' not in os.environ,
argparser.add_argument(
'--run-scale',
# Set it to the env RUN_SCALE if it isn't blank, otherwise set it to 0.25
default=os.environ['RUN_SCALE'] if 'RUN_SCALE' in os.environ and os.environ['RUN_SCALE'] != '' else 0.25, # noqa: E501
type=float,
help="The scale to run the detection at, default is 0.25",
)
# argparser.add_argument(
# '--view-scale',
# # Set it to the env VIEW_SCALE if it isn't blank, otherwise set it to 0.75
# default=os.environ['VIEW_SCALE'] if 'VIEW_SCALE' in os.environ and os.environ['VIEW_SCALE'] != '' else 0.75, # noqa: E501
# type=float,
# help="The scale to view the detection at, default is 0.75",
# )
stream_source = argparser.add_mutually_exclusive_group()
# stream_source.add_argument(
# '--url',
# default=os.environ['URL'] if 'URL' in os.environ and os.environ['URL'] != '' else None, # noqa: E501
# type=str,
# help="The URL of the stream to use",
# )
stream_source.add_argument(
'--capture-device',
default=os.environ['CAPTURE_DEVICE'] if 'CAPTURE_DEVICE' in os.environ and os.environ['CAPTURE_DEVICE'] != '' else 0, # noqa: E501
type=int,
help="The capture device to use. Can also be a url."
)
notifcation_services = argparser.add_argument_group("Notification Services")
notifcation_services.add_argument(
'--ntfy-url',
default=os.environ['NTFY_URL'] if 'NTFY_URL' in os.environ and os.environ['NTFY_URL'] != '' else None, # noqa: E501
type=str,
help="The URL to send notifications to",
)
args = argparser.parse_args()
# Check if a CUDA GPU is available. If it is, set it via torch. Ff not, set it to cpu
# https://github.com/ultralytics/ultralytics/issues/3084#issuecomment-1732433168
device = "0" if torch.cuda.is_available() else "cpu"
if device == "0":
torch.cuda.set_device(0)
print("Set CUDA device")
else:
print("No CUDA device available, using CPU")
model = YOLO("yolov8n.pt")
video_capture = cv2.VideoCapture(args.capture_device)
# Eliminate lag by setting the buffer size to 1
# This makes it so that the video capture will only grab the most recent frame
# However, this means that the video may be choppy
video_capture.set(cv2.CAP_PROP_BUFFERSIZE, 1)
# Print the resolution of the video
print(
f"Video resolution: {video_capture.get(cv2.CAP_PROP_FRAME_WIDTH)}x{video_capture.get(cv2.CAP_PROP_FRAME_HEIGHT)}" # noqa: E501
)
print("Beginning video capture...")
while True:
# Grab a single frame of video
ret, frame = video_capture.read()
# Only process every other frame of video to save time
# Resize frame of video to a smaller size for faster recognition processing
run_frame = cv2.resize(frame, (0, 0), fx=args.run_scale, fy=args.run_scale)
# view_frame = cv2.resize(frame, (0, 0), fx=args.view_scale, fy=args.view_scale)
results = model(run_frame)
for r in results:
im_array = r.plot()
# Scale back up the coordinates of the locations of detected objects.
# im_array = np.multiply(im_array, 1/args.run_scale)
# print(type(im_array))
# print(im_array)
# exit()
cv2.imshow("View", im_array)
# Hit 'q' on the keyboard to quit!
if cv2.waitKey(1) & 0xFF == ord("q"):
break
# Release handle to the webcam
print("Releasing video capture")
video_capture.release()
cv2.destroyAllWindows()
main()

0
src/utils/__init__.py Normal file
View File

4
src/utils/config_utils.py Normal file
View File

@ -0,0 +1,4 @@
# def write_config():
# with open(config_path, "w") as config_file:
# json.dump(config, config_file, indent=4)

38
src/utils/notify.py Normal file
View File

@ -0,0 +1,38 @@
import datetime
import httpx
def construct_ntfy_headers(
title: str = "Object/Person Detected",
tag = "rotating_light", # https://docs.ntfy.sh/publish/#tags-emojis
priority = "default", # https://docs.ntfy.sh/publish/#message-priority
) -> (dict):
return {
'Title': title,
'Priority': priority,
'Tags': tag
}
def send_notification(
data: str,
headers: dict,
url: str
):
if url is None or data is None:
raise ValueError("url and data cannot be None")
httpx.post(url, data=data.encode('utf-8'), headers=headers)
def check_last_seen(last_seen: datetime.datetime, seconds: int = 15):
'''
Check if a time is older than a given number of seconds
If it is, return True
If last_seen is empty/null, return True
'''
if (
datetime.datetime.now() - last_seen > datetime.timedelta(seconds=seconds)
or last_seen == ""
or last_seen is None
):
return True
else:
return False