Object detection has been studied for a long time with increasingly better prediction results. Usually, large datasets are required to get good results. In this video, we will use the pre-trained YOLOv7 model and fine-tune it on our custom dataset to recognize our custom object. As I will show in the video, really good results can be achieved with as few as 16 training images.

00:00:00 Intro
00:00:21 Getting Started
00:01:23 Data Collection
00:03:28 Labelling
00:07:37 Dataset Formatting
00:12:09 Training YOLOv7 Model
00:17:40 Detecting Your Own Class
00:23:05 Outro

Links Used in This Video

Colab Notebook:
https://colab.research.google.com/drive/1h74P5OCr76KTw8GsRR2L4JFf7MfMwBXZ?usp=sharing

Pixabay:
https://pixabay.com/

Label Studio:
https://labelstud.io/

YOLOv7 Repository:
https://github.com/WongKinYiu/yolov7

YOLOv7 Open Issue:
https://github.com/WongKinYiu/yolov7/issues/1101

Stay in Touch

Medium
https://medium.com/@martin-thissen

LinkedIn
https://linkedin.com/in/mthissen135/

YouTube
Of course, feel free to subscribe to my channel! 🙂

https://patreon.com/MartinThissen (of course, financial support is completely voluntary, but I was asked for this)

I hope this video helped you to achieve awesome object detection results 🙂 If so, feel free to share your success story in the comment section, but also if you have questions or ideas. Any like or subscription is very much appreciated 🙏🏻

source