Socket communication thread

thread1()

Pathway to receive photos and bounding box from the detection camera

• parameter: client_socket, address
• return: nothing

thread2()

Path to send photos received from detection to notice application

• parameter: client_socket, address
• return: nothing

thread3()

Path to send calculated height to notice application

• parameter: client_socket, address
• return: nothing

thread4()

Pathway to receive coordinates from tracking camera

• parameter: client_socket, address
• return: nothing

thread5()

Path through which the integrated application receives information from the tracking camera

• parameter: client_socket, address
• return: nothing

thread6()

Path to receive photos from tracking camera

• parameter: client_socket, address
• return: nothing

thread7()

Path through which the integrated application receives tracking image from the tracking camera

• parameter: client_socket, address
• return: nothing

---

RootNet

root_demo()

Function to run RootNet

• parameter : bounding box, image's byte array
• return : pose_demo(bbox_list, float(depth), img_array)

---

PoseNet

pose_demo()

Function to run PoseNet

• parameter : boudning box, root depth, image's byte array
• return : calculated person's height

PAR - Pedestrian Attribute Recognition

PAR(img)

Function to implement attribute recognition for detected person

• parameter : Image of a person cut with a bounding box
• return : attribute list of 3 most likely labels

get_reload_weight(model_path, model)

Function to setting pretrained PAR model

• parameter : pretrained model path, model (resnet50+BaseClassifier)
• return : model

get_labels(dataset='PA100k')

Function to get labels(person's attribute) about PA100k dataset

• parameter : dataset
• return : list of labels (PA100k has 26 labels)