-
Prepare your dataset and label them in YOLO format using LabelImg. Once done, zip all the images and their corresponding label files as
images.zip
. -
Create a folder named
yolov3
on Google Drive and upload theimages.zip
file inside it. The directory structure should look something like the following:
yolov3
|__images.zip
|__ *.jpg (all image files)
|__ *.txt (all annotation files)
-
Clone the repository and upload the
YOLOv3_Custom_Object_Detection.ipynb
notebook on Google Colab. -
Run the cells one-by-one by following instructions as stated in the notebook. For detailed explanation, refer the following document.
-
Once the training is completed, download the following files from the
yolov3
folder saved on Google Drive, onto your local machine.yolov3_training_last.weights
yolov3_testing.cfg
classes.txt
-
Copy the downloaded files and save them inside the repository you had cloned on your local machine.
-
Create a new folder named
test_images
inside the repository and save some test images inside it which you'd like to test the model on. -
Open the
Object_Detection.py
file inside the repository and edit Line 17 by replacing<your_test_image>
with the name of the image file you want to the test. -
Run the command:
python Object_Detection.py
.
- Read the Medium blog for step-by-step implementation.