https://github.com/aliemad5/yolo-keras-object-detection-model

This project runs real-time object detection with YOLOv11 and OpenCV. It captures webcam frames, detects objects, and displays bounding boxes with labels. It uses my custom Keras CNN for training. See details about the CNN at the website below. This code is for view and test only. If you want to use it in real projects read the LICENSE.
https://github.com/aliemad5/yolo-keras-object-detection-model

ai image keras object-detection opencv python tensorflow yolov11

Last synced: 2 months ago
JSON representation

This project runs real-time object detection with YOLOv11 and OpenCV. It captures webcam frames, detects objects, and displays bounding boxes with labels. It uses my custom Keras CNN for training. See details about the CNN at the website below. This code is for view and test only. If you want to use it in real projects read the LICENSE.

• Host: GitHub
• URL: https://github.com/aliemad5/yolo-keras-object-detection-model
• Owner: aliemad5
• License: other
• Created: 2025-09-11T13:26:21.000Z (10 months ago)
• Default Branch: main
• Last Pushed: 2025-10-04T08:01:56.000Z (9 months ago)
• Last Synced: 2025-10-04T10:06:50.900Z (9 months ago)
• Topics: ai, image, keras, object-detection, opencv, python, tensorflow, yolov11
• Homepage: https://github.com/aliemad5/Keras-custom-image-deep-learning-code?tab=readme-ov-file
• Size: 80.1 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          

Author: Ali Emad Elsamanoudy

Email:ali.elsamanoudy623@gmail.com

## License

Copyright (c) 2025 Ali Emad Elsamanoudy

[MIT License](./LICENSE) — **CREDIT REQUIRED. DO NOT IGNORE.**

## What you can do

## Requirements

- Have Google Drive App

- Have The Keras training model [Custom Keras Model Repo](https://github.com/aliemad5/Keras-custom-image-deep-learning-code)

- Have Pycharm or VScode

- Have a working Webcam

## Instructions

## Imports

```python

from ultralytics import YOLO

from tensorflow import keras

import cv2

import numpy as np

```

## Load YOLO model

```python

yolo_model = YOLO("yolov11n.pt")

```

## Load custom model

This project uses a pre-trained Keras model saved as **mykeras.h5**. The full training code, data preprocessing, and experiments for this model are explained in another repository: [Custom Keras Model Repo](https://github.com/aliemad5/Keras-custom-image-deep-learning-code)

```python

keras_model = keras.models.load_model("mykeras.h5")

```

# Open Webcam

```python

video = cv2.VideoCapture(0)

```

## Main loop

```python

while video.isOpened(): 

    ret, frame = video.read()

    if not ret:

        break

    results = yolo_model.predict(frame, verbose=False)

    boxes = results[0].boxes.xyxy.cpu().numpy().astype(int)

    if len(boxes) > 0:

        for x1, y1, x2, y2 in boxes:

            cropped = frame[y1:y2, x1:x2]

            if cropped.size == 0: 

                continue

            cropped_rgb = cv2.cvtColor(cropped, cv2.COLOR_BGR2RGB)

            resized = cv2.resize(cropped_rgb, (300, 300))

            expanded = np.expand_dims(resized, axis=0) / 255.0  # normalize

            prediction = keras_model.predict(expanded, verbose=0)

            class_id = np.argmax(prediction)

            cv2.rectangle(frame, (x1, y1), (x2, y2), (255, 255, 0), 3)

            cv2.putText(frame,

                        f"Class: {class_id}",

                        (x1, y1 - 10),

                        cv2.FONT_HERSHEY_SIMPLEX,

                        0.6,

                        (255, 0, 0),

                        2)

```

## Show output and exit

```python

    cv2.imshow("YOLO + Keras Project", frame)

   

    if cv2.waitKey(1) & 0xFF == ord("q"):

        break

video.release()

cv2.destroyAllWindows()

```