launcher.py

# ailia MODELS launcher

import glob
import os
import shutil
import subprocess
import sys
# for macOS, please install "brew install python-tk@3.9"
import tkinter as tk
import tkinter.filedialog
from tkinter import ttk

import ailia
import cv2
import numpy
from PIL import Image, ImageTk

sys.path.append('./util')
from image_utils import imread  # noqa: E402
from arg_utils import get_base_parser, update_parser  # noqa: E402

# ======================
# Arguemnt Parser Config
# ======================
parser = get_base_parser('ailia MODELS launcher', None, None)
args = update_parser(parser)

# ======================
# Global settings
# ======================

input_index = 0
output_index = 0
env_index = args.env_id
model_index = 0

# ======================
# Model search
# ======================

IGNORE_LIST = [
    "commercial_model", "validation", ".git", "log", "prnet", "bert", "neural_rendering",
    "illustration2vec", "etl", "vggface2", "anomaly_detection", "natural_language_processing", "audio_processing"
]

def get_model_list():
    global model_index

    file_list = []
    for current, subfolders, subfiles in os.walk("./"):
        file_list.append(current)

    file_list.sort()

    model_list = []
    category_list = {}
    model_exist = {}
    for current in file_list:
        current = current.replace("\\", "/")
        files = current.split("/")
        if len(files) == 3:
            if (files[1] in IGNORE_LIST) or (files[2] in IGNORE_LIST):
                continue
            if files[2] in model_exist:
                continue
            script = "./"+files[1]+"/"+files[2]+"/"+files[2]+".py"
            if os.path.exists(script):
                if not(files[1] in category_list):
                    category_list[files[1]] = len(category_list)
                category_id = category_list[files[1]]
                model_list.append({
                    "category": files[1],
                    "category_id": category_id,
                    "model": files[2],
                })
                model_exist[files[2]] = True


    model_name_list = []
    for i in range(len(model_list)):
        model_name_list.append(""+model_list[i]["category"]+" : "+model_list[i]["model"])
        if model_list[i]["model"]=="yolox":
            model_index = i

    return model_list, model_name_list, len(category_list)


# ======================
# Environment
# ======================

def get_input_list():
    if args.debug:
        return ["Camera:0"]

    index = 0
    inputs = []
    while True:
        cap = cv2.VideoCapture(index)
        if cap.isOpened():
            inputs.append("Camera:"+str(index))
        else:
            break
        index=index+1
        cap.release()
    return inputs

def input_changed(event):
    global input_index
    selection = event.widget.curselection()
    if selection:
        input_index = selection[0]
    else:
        input_index = 0   
    #print("input",input_index)

def get_output_list():
    return ["Display:0"]

def output_changed(event):
    global output_index
    selection = event.widget.curselection()
    if selection:
        output_index = selection[0]
    else:
        output_index = 0   
    #print("output",output_index)

def get_env_list():
    env_list = []
    for env in ailia.get_environment_list():
        env_list.append(env.name)
    return env_list  

def environment_changed(event):
    global env_index
    selection = event.widget.curselection()
    if selection:
        env_index = selection[0]
    else:
        env_index = 0
    #print("env",env_index)

# ======================
# Change model
# ======================


def model_changed(event):
    global model_index
    selection = event.widget.curselection()
    if selection:
        model_index = selection[0]
    else:
        model_index = 0
    load_detail(model_index)

def create_photo_image(path,w=320,h=240):
    image_bgr = imread(path)
    #image_bgr = cv2.resize(image_bgr,(w,h))
    image_rgb = cv2.cvtColor(image_bgr, cv2.COLOR_BGR2RGB) # imreadはBGRなのでRGBに変換
    image_pil = Image.fromarray(image_rgb) # RGBからPILフォーマットへ変換
    image_pil.thumbnail((w,h), Image.LANCZOS)
    image_tk  = ImageTk.PhotoImage(image_pil) # ImageTkフォーマットへ変換
    return image_tk

def load_image(path):
    if not os.path.isfile(path):
        return
    global canvas, canvas_item, image_tk
    image_tk = create_photo_image(path)
    if canvas_item == None:
        canvas_item = canvas.create_image(0, 0, image=image_tk, anchor=tk.NW)
    else:
        canvas.itemconfig(canvas_item,image=image_tk)

def load_detail(index):
    model_request = model_list[index]

    base_path =  "./"+model_request["category"]+"/"+model_request["model"]+"/"

    image_exist = False
    for ext in [".jpg",".png"]:
        image_path = base_path + "output" + ext
        if os.path.exists(image_path):
            load_image(image_path)
            image_exist = True
            break

    if not image_exist:
        files = glob.glob(base_path+"*.jpg")
        files.extend(glob.glob(base_path+"*.png"))
        for image_path in files:
            if os.path.exists(image_path):
                load_image(image_path)
                break

    global textModelDetail

    f = open(base_path+"README.md", encoding="utf8")
    text = f.readlines()
    text = text[0].replace("# ","")
    text = text.replace("\r","")
    text = text.replace("\n","")
    text = text[:40]
    textModelDetail.set(text)


# ======================
# Run model
# ======================

proc = None

def open_model(model):
    global proc

    if not (proc==None):
        proc.kill()
        proc=None

    model_request = model_list[model_index]

    dir = "./"+model["category"]+"/"+model["model"]+"/"
    cmd = sys.executable

    args_dict = vars(args)

    if "Camera:" in input_list[input_index]:
        video_name = input_index
    else:
        video_name = input_list[input_index]
    
    if "Display:" in output_list[output_index]:
        save_name = "temp.png"
        save_path = "./"+model_request["category"]+"/"+model_request["model"]+"/"+"temp.png"
    else:
        save_name = output_list[output_index]
        save_path = output_list[output_index]

    if not(("natural_language_processing" == model["category"]) or ("audio_processing" == model["category"])):
        if ".png" in str(video_name) or ".jpg" in str(video_name):
            if "video" in args_dict:
                del args_dict["video"]
            args_dict["input"]=video_name
            args_dict["savepath"]=save_name
        else:
            args_dict["video"]=video_name
            if not ("Display:" in output_list[output_index]):
                args_dict["savepath"]=save_name

    args_dict["env_id"]=env_index

    options = []
    for key in args_dict:
        if key=="ftype":
            continue
        if args_dict[key] is not None:
            if args_dict[key] is True:
                options.append("--"+key)
            elif args_dict[key] is False:
                continue
            else:
                options.append("--"+key)
                options.append(str(args_dict[key]))
    
    cmd = [cmd, model["model"]+".py"] + options
    print(" ".join(cmd))

    if not ("video" in args_dict):
        subprocess.check_call(cmd, cwd=dir, shell=False)
        load_image(save_path)
    else:
        proc = subprocess.Popen(cmd, cwd=dir)
        try:
            outs, errs = proc.communicate(timeout=1)
        except subprocess.TimeoutExpired:
            pass

def run_button_clicked():
    global model_list
    model_request=model_list[int(model_index)]
    open_model(model_request)

def stop_button_clicked():
    global proc

    if not (proc==None):
        proc.kill()
        proc=None

# ======================
# IP Camera
# ======================

camerasWindow = None

def input_camera_dialog():
    global camerasWindow

    if camerasWindow != None and camerasWindow.winfo_exists():
        return

    camerasWindow = tk.Toplevel()
    camerasWindow.title("IP Camera Settings")
    camerasWindow.geometry("300x300")

    frame = ttk.Frame(camerasWindow)
    frame.pack(padx=10,pady=10)

    textOptions = tk.StringVar(frame)
    textOptions.set("IP camera address (rtsp://*)")
    labelOptions = tk.Label(frame, textvariable=textOptions)
    labelOptions.grid(row=0, column=0, sticky=tk.NW)

    global cameraEntry
    cameraEntry = tkinter.Entry(frame, width=20)
    cameraEntry.insert(tkinter.END, "")
    cameraEntry.grid(row=1, column=0, sticky=tk.NW, rowspan=1)

    setCamerasSettingsText = tk.StringVar(frame)
    setCamerasSettingsText.set("OK")
    buttonSetCamerasSettings = tk.Button(frame, textvariable=setCamerasSettingsText, command=input_camera_dialog_close, width=14)
    buttonSetCamerasSettings.grid(row=4, column=0, sticky=tk.NW)

def input_camera_dialog_close():
    global cameraEntry
    global listsInput, ListboxInput, input_index
    ip_camera = cameraEntry.get()
    if ip_camera != "":
        input_list.append(ip_camera)
        listsInput.set(input_list)
        ListboxInput.select_clear(input_index)
        input_index = len(input_list)-1
        ListboxInput.select_set(input_index)

    global camerasWindow
    camerasWindow.destroy()
    camerasWindow = None

# ======================
# Select file
# ======================

def input_file_dialog():
    global listsInput, ListboxInput, input_index
    fTyp = [("Image File or Video File", "*")]
    iDir = os.path.abspath(os.path.dirname(__file__))
    file_name = tk.filedialog.askopenfilename(filetypes=fTyp, initialdir=iDir)
    if len(file_name) != 0:
        input_list.append(file_name)
        listsInput.set(input_list)
        ListboxInput.select_clear(input_index)
        input_index = len(input_list)-1
        ListboxInput.select_set(input_index)

def output_file_dialog():
    global listsOutput, ListboxOutput, output_index
    fTyp = [("Image File or Video File", "*")]
    iDir = os.path.abspath(os.path.dirname(__file__))
    file_name = tk.filedialog.asksaveasfilename(filetypes=fTyp, initialdir=iDir)
    if len(file_name) != 0:
        output_list.append(file_name)
        listsOutput.set(output_list)
        ListboxOutput.select_clear(output_index)
        output_index = len(output_list)-1
        ListboxOutput.select_set(output_index)

# ======================
# GUI
# ======================

canvas_item = None

def main():
    global ListboxModel, textModelDetail
    global model_list, model_request, model_loading_cnt, invalidate_quit_cnt
    global canvas
    global inputFile, listsInput, input_list, ListboxInput
    global outputFile, listsOutput, output_list, ListboxOutput

    model_list, model_name_list, category_cnt = get_model_list()

    # rootメインウィンドウの設定
    root = tk.Tk()
    root.title("ailia MODELS")
    root.geometry("1200x600")

    # メインフレームの作成と設置
    frame = ttk.Frame(root)
    frame.pack(padx=20,pady=10)

    # Listboxの選択肢
    env_list = get_env_list()
    input_list = get_input_list()
    output_list = get_output_list()

    lists = tk.StringVar(value=model_name_list)
    listsInput = tk.StringVar(value=input_list)
    listsOutput = tk.StringVar(value=output_list)
    listEnvironment =tk.StringVar(value=env_list)

    # 各種ウィジェットの作成
    ListboxModel = tk.Listbox(frame, listvariable=lists, width=40, height=30, selectmode=tk.BROWSE, exportselection=False)
    ListboxInput = tk.Listbox(frame, listvariable=listsInput, width=40, height=4, selectmode=tk.BROWSE, exportselection=False)
    ListboxOutput = tk.Listbox(frame, listvariable=listsOutput, width=40, height=4, selectmode=tk.BROWSE, exportselection=False)
    ListboxEnvironment = tk.Listbox(frame, listvariable=listEnvironment, width=40, height=4, selectmode=tk.BROWSE, exportselection=False)

    ListboxModel.bind("<<ListboxSelect>>", model_changed)
    ListboxInput.bind("<<ListboxSelect>>", input_changed)
    ListboxOutput.bind("<<ListboxSelect>>", output_changed)
    ListboxEnvironment.bind("<<ListboxSelect>>", environment_changed)

    ListboxModel.select_set(model_index)
    ListboxInput.select_set(input_index)
    ListboxOutput.select_set(output_index)
    ListboxEnvironment.select_set(env_index)

    # スクロールバーの作成
    scrollbar = tk.Scrollbar(frame, orient=tk.VERTICAL, command=ListboxModel.yview)

    # スクロールバーをListboxに反映
    ListboxModel["yscrollcommand"] = scrollbar.set

    # StringVarのインスタンスを格納する変数textの設定
    textRun = tk.StringVar(frame)
    textRun.set("Run model")

    textStop = tk.StringVar(frame)
    textStop.set("Stop model")

    textInputFile = tk.StringVar(frame)
    textInputFile.set("Add input file")

    textInputCamera = tk.StringVar(frame)
    textInputCamera.set("Add ip camera")

    textOutputFile = tk.StringVar(frame)
    textOutputFile.set("Add output file")

    textModel = tk.StringVar(frame)
    textModel.set("Models")

    textInput = tk.StringVar(frame)
    textInput.set("Input")

    textOutput = tk.StringVar(frame)
    textOutput.set("Output")

    textEnvironment = tk.StringVar(frame)
    textEnvironment.set("Environment")

    textModelDetail = tk.StringVar(frame)
    textModelDetail.set("ModelDetail")

    # 各種ウィジェットの作成
    labelModel = tk.Label(frame, textvariable=textModel)
    labelInput = tk.Label(frame, textvariable=textInput)
    labelOutput = tk.Label(frame, textvariable=textOutput)
    labelEnvironment = tk.Label(frame, textvariable=textEnvironment)
    labelModelDetail = tk.Label(frame, textvariable=textModelDetail)

    buttonRun = tk.Button(frame, textvariable=textRun, command=run_button_clicked, width=14)
    buttonStop = tk.Button(frame, textvariable=textStop, command=stop_button_clicked, width=14)
    buttonInputFile = tk.Button(frame, textvariable=textInputFile, command=input_file_dialog, width=14)
    buttonInputCamera = tk.Button(frame, textvariable=textInputCamera, command=input_camera_dialog, width=14)
    buttonOutputFile = tk.Button(frame, textvariable=textOutputFile, command=output_file_dialog, width=14)

    canvas = tk.Canvas(frame, bg="black", width=320, height=240)
    canvas.place(x=0, y=0)
    load_detail(model_index)

    logo = tk.Canvas(frame, bg="black", width=320, height=124)
    logo.place(x=0, y=0)
    global logo_img
    logo_img = create_photo_image("ailia-models.png",320,124)
    logo_item = logo.create_image(0, 0, image=logo_img, anchor=tk.NW)

    # 各種ウィジェットの設置
    labelModel.grid(row=0, column=0, sticky=tk.NW, rowspan=12)
    ListboxModel.grid(row=1, column=0, sticky=tk.NW, rowspan=12)
    scrollbar.grid(row=1, column=1, sticky=(tk.N, tk.S), rowspan=12)

    labelModelDetail.grid(row=0, column=2, sticky=tk.NW)
    canvas.grid(row=1, column=2, sticky=tk.NW, rowspan=4, columnspan=2)

    labelInput.grid(row=5, column=2, sticky=tk.NW, columnspan=2)
    ListboxInput.grid(row=6, column=2, sticky=tk.NW, columnspan=2)
    buttonInputFile.grid(row=7, column=2, sticky=tk.NW)
    buttonInputCamera.grid(row=7, column=3, sticky=tk.NW)

    labelOutput.grid(row=8, column=2, sticky=tk.NW)
    ListboxOutput.grid(row=9, column=2, sticky=tk.NW, columnspan=2)
    buttonOutputFile.grid(row=10, column=2, sticky=tk.NW)

    labelEnvironment.grid(row=0, column=4, sticky=tk.NW, columnspan=2)
    ListboxEnvironment.grid(row=1, column=4, sticky=tk.NW, columnspan=2)

    buttonRun.grid(row=3, column=4, sticky=tk.NW)
    buttonStop.grid(row=3, column=5, sticky=tk.NW)

    logo.grid(row=2, column=4, sticky=tk.NW, columnspan=2, rowspan=1)

    # メインフレームの作成と設置
    frame = ttk.Frame(root)
    frame.pack(padx=20, pady=10)

    root.mainloop()

if __name__ == '__main__':
    main()