In this tutorial, we will find steps to run Yolo Object detection on Google Colab with GPU. This post gives you a general idea about detection using pre-trained model on both image and video.
Lets start by creating Python notebook with Google Colab, all you need is a Google account and to click the link https://colab.research.google.com. If you aren’t aware of Google Colab environment, you should watch the video Get started with Google Colaboratory. Hardware specification available in Google Colab: > Number of CPU: 2 (32bit, 64bit) > CPU: Intel(R) Xeon(R) CPU @ 2.30GHz > CPU Cores: 1 > RAM: ~12GB (Actual 6-8GB) > Disk: ~68GB (Available ~36GB) > GPU: 1xTesla k80, 12GB > CUDA Version: 10.1 > CUDA Cores: 4992
You only look once (YOLO) is a state-of-the-art, real-time object detection system. You will find a full explanation about yolo on https://pjreddie.com/darknet/yolo/. In this post, we will use YOLOv3 weights trained on COCO dataset. COCO dataset contains 80 classes. You can read all classes here. To read more about COCO dataset go to this link.