config.sample

# AI framework - one of "AWS", "TF" and "SH" for Amazon Rekognition, TensorFlow and SightHound resp.
# MODEL_SYSTEM = "AWS"
MODEL_SYSTEM = "TF"
# MODEL_SYSTEM = "SH"


DEBUG = False

# Push notifications
PUSH_MQTT_ALERT = False #Send an alert back to BI camera with same name plus "_alert" suffix

PUSH_PUSHBULLET_NOTIFICATION = False
PUSHBULLET_API_KEY = ""

PUSH_TELEGRAM_ALERT = True
TELEGRAM_API_TOKEN = ""
TELEGRAM_CHAT_ID = 

# Note that even if set to True, original images are only saved if they do not have any identified objects
SAVE_ORIGINAL_IMAGE = True

# Processed images are those that have been processed AND have objects that we are looking for
# Image name will include the objects recognised
SAVE_PROCESSED_IMAGE = True
TAG_IMAGES = True
BOX_NON_TRIGGER_OBJECTS = True

# Dictionary list of box corners by camera that are to be ignored - NOT YET IMPLEMENTED
IGNORE_CAM_REGIONS = {"Driveway" : [[(600, 0), (800, 90)]]}

# Object labels that we are looking for. These should match the labels used by the model being used
# e.g. AWS Rekognition may have different labels to Coco data set (as used in the TensorFlow model) etc.
DETECT_OBJECTS_DEFAULT = "Person"
DETECT_OBJECTS_BY_CAMERA = {"Driveway" : "Person Car Lorry Van Truck Bike Motorcycle",
                            "Side" : "Person Bike Motorcycle", 
                            "GardenMain" : "Person"}

#Confidence threshold is a percentage out of 100
CONFIDENCE_THRESHOLD = 70.0

# TensorFlow Model
# TENSORFLOW_MODEL_DIRECTORY = "./ssd_mobilenet_v1_coco_2017_11_17"
TENSORFLOW_MODEL_DIRECTORY = "./faster_rcnn_inception_v2_coco_2018_01_28"
FROZEN_INFERENCE_GRAPH = "frozen_inference_graph.pb"


# Sighthound Cloud
SIGHTHOUND_DETECT_TYPES = "type=all&faceOption=gender"
SIGHTHOUND_CLOUD_TOKEN = ""

# Blue Iris Server (can use IP address in BI_SERVER)
BI_SERVER = ""
BI_PORT = 80
BI_PROTOCOL = "http"
BI_USER = ""
BI_PW = ""

# Scale is any number > 0. Quality is 1 to 100%
BI_IMAGE_SCALE = 1.0
BI_IMAGE_QUALITY = 100

# This parameter basically says to treat camera names with anything before last "_" seperator
# as cloned cameras e.g. "Driveway" and "Driveway_Cloned" would be treated as the same camera
# (this helps by sharing stationary objects list between the cloned cameras)
BI_CLONED_CAMERA_COMMON_NAMING = False

# MQTT Broker
MQTT_SERVER = ""
MQTT_USER = ""
MQTT_PW = ""
MQTT_LISTEN_TOPIC = "BlueIris/alert"
MQTT_PUBLISH_TOPIC = "BlueIris/admin"

IMAGE_SAVE_PATH = "./images"