Python

Bot for analysis of the image by photo, based on yolov11

Сделано на базе YOLOv11. Обнаруживает и подсчитывает объекты на фото - людей, авто, животных, и тд. Сделано было чисто ради изучения и не несёт какой-то пользы, НО с помощью YOLO можно делать много реально интересных вещей при должных знаниях.
YOLO (You Only Look Once) — система обнаружения объектов в реальном времени, разработанная Джозефом Редмоном и Али Фархади в 2015 году. Название означает «вы смотрите только один раз» — алгоритм обрабатывает всё изображение за один проход, в отличие от традиционных методов, которые анализируют каждый объект по отдельности.

Как работает:​

Запускаете, открываете любую картину, выдаёт готовый результат. Результаты хранятся в корневой папке output.
​

Так же сделал бота для ознакомления, чтобы вам ничего не устанавливать.​

Telegram: ANALISOBJ613246_BOT - заходите, нажимаете /start, отправляете картинку
​

Код

1. Для начала нужен Python, я использовал Python 3.10
2. Нужно установить зависимости - pip install ultralytics tkinterdnd2 tkinter torch pillow
3. Сам код:

import tkinter as tk
from tkinter import filedialog, messagebox, ttk
from PIL import Image, ImageTk, ImageGrab
import os
import time
import uuid
import sys
import re

try:
from tkinterdnd2 import DND_FILES, TkinterDnD
DND_AVAILABLE = True
except Exception:
DND_AVAILABLE = False

try:
from ultralytics import YOLO
HAS_ULTRALYTICS = True
except Exception:
YOLO = None
HAS_ULTRALYTICS = False

try:
import torch
HAS_TORCH = True
except Exception:
HAS_TORCH = False

yolo_model = None
MODEL_NAME = "yolo11n.pt"
OUTPUT_DIR = "output"
ALLOWED_EXT = ('.jpg', '.jpeg', '.png', '.bmp', '.gif', '.webp')

current_output_path = None
display_original_pil = None
zoom_factor = 1.0

ANIMAL_CLASSES = {"bird","cat","dog","bear","cow"}

def ensure_output_dir():
os.makedirs(OUTPUT_DIR, exist_ok=True)
testfile = os.path.join(OUTPUT_DIR, ".write_test")
try:
with open(testfile, "w", encoding="utf-8") as f:
f.write("ok")
os.remove(testfile)
except Exception as e:
raise RuntimeError(f"")

def lazy_load_model():
global yolo_model
if yolo_model is not None:
return
if not HAS_ULTRALYTICS:
raise RuntimeError("")
try:
yolo_model = YOLO(MODEL_NAME)
except Exception as e:
raise RuntimeError(f"ошибка загрузки модели '{MODEL_NAME}': {e}")

def analyze_image(image_path):
ensure_output_dir()
lazy_load_model()

try:
img = Image.open(image_path).convert("RGB")
except Exception as e:
raise RuntimeError(f"ошибка при загрузке '{image_path}': {e}")

try:
results = yolo_model(image_path)
except Exception as e:
raise RuntimeError(f"ошибка модели: {e}")

COCO_CLASSES = getattr(yolo_model, "names", None) or {}
people_count = 0
vehicle_count = 0
animals = {}

for result in results:
boxes = getattr(result, "boxes", None)
if boxes is None:
continue
for box in boxes:
try:
cls_id = int(box.cls[0].item())
except Exception:
continue
cls_name = COCO_CLASSES.get(cls_id, str(cls_id)).lower()
if cls_name == "person":
people_count += 1
elif cls_name in ("car","truck","bus","motorcycle"):
vehicle_count += 1
elif cls_name in ANIMAL_CLASSES:
animals[cls_name] = animals.get(cls_name, 0) + 1

timestamp = time.strftime("%Y%m%d_%H%M%S")
uniq = uuid.uuid4().hex[:8]
save_path = os.path.join(OUTPUT_DIR, f"result_{timestamp}_{uniq}.jpg")
try:
results[0].save(filename=save_path)
except Exception:
img.save(save_path)

return save_path, people_count, vehicle_count, animals

def update_display_image():
global display_original_pil, zoom_factor, current_display_photo
if display_original_pil is None:
return
max_w = max(100, app.winfo_width() - 60)
max_h = max(100, app.winfo_height() - 160)
w, h = display_original_pil.size
tw = int(w * zoom_factor)
th = int(h * zoom_factor)
if tw > max_w or th > max_h:
scale = min(max_w / tw, max_h / th)
tw = max(1, int(tw * scale))
th = max(1, int(th * scale))
img_resized = display_original_pil.resize((tw, th), Image.LANCZOS)
photo = ImageTk.PhotoImage(img_resized)
image_label.config(image=photo)
image_label.image = photo
current_display_photo = photo

def display_image(path):
global display_original_pil, zoom_factor
try:
img = Image.open(path).convert("RGB")
except Exception as e:
messagebox.showerror("ошибка", f"не удалось открыть файфл: {e}")
return
display_original_pil = img
zoom_factor = 1.0
update_display_image()

def process_image(path):
global current_output_path
info_text.set("анализ")
app.update_idletasks()
try:
save_path, people_count, vehicle_count, animals = analyze_image(path)
except Exception as e:
messagebox.showerror("ошибка при анализе", str(e))
info_text.set("ошибка")
return
current_output_path = save_path
display_image(save_path)
animals_text = ", ".join(f"{k}({v})" for k,v in animals.items()) if animals else "-"
info_text.set(f"людей: {people_count} авто: {vehicle_count} животные: {animals_text}")
print(f"сохранено в {save_path}")

def open_file():
path = filedialog.askopenfilename(filetypes=[("Images", "*.jpg;*.jpeg;*.png;*.bmp;*.gif;*.webp")])
if path:
process_image(path)

def parse_dnd_paths(data_str):
parts = re.findall(r'{([^}]*)}', data_str)
if not parts:
parts = data_str.split()
return parts

def paste_clipboard(event=None):
try:
cb = ImageGrab.grabclipboard()
except Exception as e:
messagebox.showerror("ошибка", f"буфер обмена пуст {e}")
return
if isinstance(cb, Image.Image):
try:
ensure_output_dir()
temp_path = os.path.join(OUTPUT_DIR, f"clipboard_{int(time.time()*1000)}.png")
cb.save(temp_path)
process_image(temp_path)
except Exception as e:
messagebox.showerror("ошибка", f"не удалось сохранить изображение из буфера: {e}")
elif isinstance(cb, list) and cb:
first = cb[0]
if isinstance(first, str) and os.path.isfile(first):
process_image(first)
else:
messagebox.showinfo("ошибка")
elif isinstance(cb, str) and os.path.isfile(cb):
process_image(cb)
else:
try:
txt = app.clipboard_get()
if txt and os.path.isfile(txt.strip()):
process_image(txt.strip())
return
except Exception:
pass
messagebox.showinfo("буфер обмена пуст")

def drop_file(event):
paths = parse_dnd_paths(event.data)
if not paths:
return
first = paths[0]
if os.path.isfile(first) and first.lower().endswith(ALLOWED_EXT):
process_image(first)
else:
messagebox.showinfo("Drag&Drop", "файл не найден/не является поддерживаемым форматом")

def close_image(event=None):
global display_original_pil, current_output_path
image_label.config(image="")
image_label.image = None
display_original_pil = None
current_output_path = None
info_text.set("закрыто")

def on_mouse_wheel(event):
global zoom_factor
delta = 0
if hasattr(event, "delta") and event.delta:
delta = event.delta
elif event.num == 4:
delta = 120
elif event.num == 5:
delta = -120
if delta > 0:
zoom_factor *= 1.1
else:
zoom_factor /= 1.1
zoom_factor = max(0.1, min(10.0, zoom_factor))
update_display_image()

def self_check():
msgs = []
if not HAS_ULTRALYTICS:
msgs.append("ultralytics не установлен - pip install -U ultralytics")
if not HAS_TORCH:
msgs.append("torch не найден")
try:
ensure_output_dir()
except Exception as e:
msgs.append(str(e))
if msgs:
messagebox.showwarning("проблемы:\n\n" + "\n".join(msgs))
return False
try:
lazy_load_model()
except Exception as e:
messagebox.showwarning("модель YOLOv11", f"не загрузилась новая модель '{MODEL_NAME}':\n{e}\n")
return True

#интерфейс
if DND_AVAILABLE:
app = TkinterDnD.Tk()
else:
app = tk.Tk()

app.title("анализ объектов YOLOv11")
app.configure(bg="#2b2b2b")
try:
app.state("zoomed")
except Exception:
pass

style = ttk.Style(app)
style.configure("TButton", font=("Arial", 12))

top_frame = tk.Frame(app, bg="#2b2b2b")
top_frame.pack(fill="x", pady=8)

open_btn = ttk.Button(top_frame, text="открыть картинку", command=open_file)
open_btn.pack(side="left", padx=10)

close_btn = ttk.Button(top_frame, text="закрыть", command=close_image)
close_btn.pack(side="left", padx=10)

info_text = tk.StringVar(value="загрузить, через drag&drop или выбрать из папки")
info_label = tk.Label(top_frame, textvariable=info_text, fg="white", bg="#2b2b2b", font=("Arial", 12))
info_label.pack(side="left", padx=20)

canvas_frame = tk.Frame(app, bg="#1f1f1f")
canvas_frame.pack(fill="both", expand=True, padx=10, pady=10)

image_label = tk.Label(canvas_frame, bg="#1f1f1f")
image_label.pack(expand=True)

app.bind_all("<Control-v>", paste_clipboard)
app.bind_all("<Control-V>", paste_clipboard)
app.bind_all("<Control-w>", close_image)
app.bind_all("<Control-W>", close_image)

if DND_AVAILABLE:
try:
image_label.drop_target_register(DND_FILES)
image_label.dnd_bind("<<Drop>>", drop_file)
except Exception as e:
print("ошибка DND", e)

_ok = self_check()
if not _ok:
print("")

app.mainloop()

Python

import tkinter as tk

from tkinter import filedialog, messagebox, ttk

from PIL import Image, ImageTk, ImageGrab

import os

import time

import uuid

import sys

import re



try:

    from tkinterdnd2 import DND_FILES, TkinterDnD

    DND_AVAILABLE = True

except Exception:

    DND_AVAILABLE = False



try:

    from ultralytics import YOLO

    HAS_ULTRALYTICS = True

except Exception:

    YOLO = None

    HAS_ULTRALYTICS = False



try:

    import torch

    HAS_TORCH = True

except Exception:

    HAS_TORCH = False



yolo_model = None

MODEL_NAME = "yolo11n.pt"

OUTPUT_DIR = "output"

ALLOWED_EXT = ('.jpg', '.jpeg', '.png', '.bmp', '.gif', '.webp')



current_output_path = None

display_original_pil = None

zoom_factor = 1.0



ANIMAL_CLASSES = {"bird","cat","dog","bear","cow"}



def ensure_output_dir():

    os.makedirs(OUTPUT_DIR, exist_ok=True)

    testfile = os.path.join(OUTPUT_DIR, ".write_test")

    try:

        with open(testfile, "w", encoding="utf-8") as f:

            f.write("ok")

        os.remove(testfile)

    except Exception as e:

        raise RuntimeError(f"")



def lazy_load_model():

    global yolo_model

    if yolo_model is not None:

        return

    if not HAS_ULTRALYTICS:

        raise RuntimeError("")

    try:

        yolo_model = YOLO(MODEL_NAME)

    except Exception as e:

        raise RuntimeError(f"ошибка загрузки модели '{MODEL_NAME}': {e}")



def analyze_image(image_path):

    ensure_output_dir()

    lazy_load_model()



    try:

        img = Image.open(image_path).convert("RGB")

    except Exception as e:

        raise RuntimeError(f"ошибка при загрузке '{image_path}': {e}")



    try:

        results = yolo_model(image_path)

    except Exception as e:

        raise RuntimeError(f"ошибка модели: {e}")



    COCO_CLASSES = getattr(yolo_model, "names", None) or {}

    people_count = 0

    vehicle_count = 0

    animals = {}



    for result in results:

        boxes = getattr(result, "boxes", None)

        if boxes is None:

            continue

        for box in boxes:

            try:

                cls_id = int(box.cls[0].item())

            except Exception:

                continue

            cls_name = COCO_CLASSES.get(cls_id, str(cls_id)).lower()

            if cls_name == "person":

                people_count += 1

            elif cls_name in ("car","truck","bus","motorcycle"):

                vehicle_count += 1

            elif cls_name in ANIMAL_CLASSES:

                animals[cls_name] = animals.get(cls_name, 0) + 1



    timestamp = time.strftime("%Y%m%d_%H%M%S")

    uniq = uuid.uuid4().hex[:8]

    save_path = os.path.join(OUTPUT_DIR, f"result_{timestamp}_{uniq}.jpg")

    try:

        results[0].save(filename=save_path)

    except Exception:

        img.save(save_path)



    return save_path, people_count, vehicle_count, animals



def update_display_image():

    global display_original_pil, zoom_factor, current_display_photo

    if display_original_pil is None:

        return

    max_w = max(100, app.winfo_width() - 60)

    max_h = max(100, app.winfo_height() - 160)

    w, h = display_original_pil.size

    tw = int(w * zoom_factor)

    th = int(h * zoom_factor)

    if tw > max_w or th > max_h:

        scale = min(max_w / tw, max_h / th)

        tw = max(1, int(tw * scale))

        th = max(1, int(th * scale))

    img_resized = display_original_pil.resize((tw, th), Image.LANCZOS)

    photo = ImageTk.PhotoImage(img_resized)

    image_label.config(image=photo)

    image_label.image = photo

    current_display_photo = photo



def display_image(path):

    global display_original_pil, zoom_factor

    try:

        img = Image.open(path).convert("RGB")

    except Exception as e:

        messagebox.showerror("ошибка", f"не удалось открыть файфл: {e}")

        return

    display_original_pil = img

    zoom_factor = 1.0

    update_display_image()



def process_image(path):

    global current_output_path

    info_text.set("анализ")

    app.update_idletasks()

    try:

        save_path, people_count, vehicle_count, animals = analyze_image(path)

    except Exception as e:

        messagebox.showerror("ошибка при анализе", str(e))

        info_text.set("ошибка")

        return

    current_output_path = save_path

    display_image(save_path)

    animals_text = ", ".join(f"{k}({v})" for k,v in animals.items()) if animals else "-"

    info_text.set(f"людей: {people_count}   авто: {vehicle_count}   животные: {animals_text}")

    print(f"сохранено в {save_path}")



def open_file():

    path = filedialog.askopenfilename(filetypes=[("Images", "*.jpg;*.jpeg;*.png;*.bmp;*.gif;*.webp")])

    if path:

        process_image(path)



def parse_dnd_paths(data_str):

    parts = re.findall(r'{([^}]*)}', data_str)

    if not parts:

        parts = data_str.split()

    return parts



def paste_clipboard(event=None):

    try:

        cb = ImageGrab.grabclipboard()

    except Exception as e:

        messagebox.showerror("ошибка", f"буфер обмена пуст {e}")

        return

    if isinstance(cb, Image.Image):

        try:

            ensure_output_dir()

            temp_path = os.path.join(OUTPUT_DIR, f"clipboard_{int(time.time()*1000)}.png")

            cb.save(temp_path)

            process_image(temp_path)

        except Exception as e:

            messagebox.showerror("ошибка", f"не удалось сохранить изображение из буфера: {e}")

    elif isinstance(cb, list) and cb:

        first = cb[0]

        if isinstance(first, str) and os.path.isfile(first):

            process_image(first)

        else:

            messagebox.showinfo("ошибка")

    elif isinstance(cb, str) and os.path.isfile(cb):

        process_image(cb)

    else:

        try:

            txt = app.clipboard_get()

            if txt and os.path.isfile(txt.strip()):

                process_image(txt.strip())

                return

        except Exception:

            pass

        messagebox.showinfo("буфер обмена пуст")



def drop_file(event):

    paths = parse_dnd_paths(event.data)

    if not paths:

        return

    first = paths[0]

    if os.path.isfile(first) and first.lower().endswith(ALLOWED_EXT):

        process_image(first)

    else:

        messagebox.showinfo("Drag&Drop", "файл не найден/не является поддерживаемым форматом")



def close_image(event=None):

    global display_original_pil, current_output_path

    image_label.config(image="")

    image_label.image = None

    display_original_pil = None

    current_output_path = None

    info_text.set("закрыто")



def on_mouse_wheel(event):

    global zoom_factor

    delta = 0

    if hasattr(event, "delta") and event.delta:

        delta = event.delta

    elif event.num == 4:

        delta = 120

    elif event.num == 5:

        delta = -120

    if delta > 0:

        zoom_factor *= 1.1

    else:

        zoom_factor /= 1.1

    zoom_factor = max(0.1, min(10.0, zoom_factor))

    update_display_image()



def self_check():

    msgs = []

    if not HAS_ULTRALYTICS:

        msgs.append("ultralytics не установлен - pip install -U ultralytics")

    if not HAS_TORCH:

        msgs.append("torch не найден")

    try:

        ensure_output_dir()

    except Exception as e:

        msgs.append(str(e))

    if msgs:

        messagebox.showwarning("проблемы:\n\n" + "\n".join(msgs))

        return False

    try:

        lazy_load_model()

    except Exception as e:

        messagebox.showwarning("модель YOLOv11", f"не загрузилась новая модель '{MODEL_NAME}':\n{e}\n")

    return True



#интерфейс

if DND_AVAILABLE:

    app = TkinterDnD.Tk()

else:

    app = tk.Tk()



app.title("анализ объектов YOLOv11")

app.configure(bg="#2b2b2b")

try:

    app.state("zoomed")

except Exception:

    pass



style = ttk.Style(app)

style.configure("TButton", font=("Arial", 12))



top_frame = tk.Frame(app, bg="#2b2b2b")

top_frame.pack(fill="x", pady=8)



open_btn = ttk.Button(top_frame, text="открыть картинку", command=open_file)

open_btn.pack(side="left", padx=10)



close_btn = ttk.Button(top_frame, text="закрыть", command=close_image)

close_btn.pack(side="left", padx=10)



info_text = tk.StringVar(value="загрузить, через drag&drop или выбрать из папки")

info_label = tk.Label(top_frame, textvariable=info_text, fg="white", bg="#2b2b2b", font=("Arial", 12))

info_label.pack(side="left", padx=20)



canvas_frame = tk.Frame(app, bg="#1f1f1f")

canvas_frame.pack(fill="both", expand=True, padx=10, pady=10)



image_label = tk.Label(canvas_frame, bg="#1f1f1f")

image_label.pack(expand=True)



app.bind_all("<Control-v>", paste_clipboard)

app.bind_all("<Control-V>", paste_clipboard)

app.bind_all("<Control-w>", close_image)

app.bind_all("<Control-W>", close_image)



if DND_AVAILABLE:

    try:

        image_label.drop_target_register(DND_FILES)

        image_label.dnd_bind("<<Drop>>", drop_file)

    except Exception as e:

        print("ошибка DND", e)



_ok = self_check()

if not _ok:

    print("")



app.mainloop()

Скрины работы