OlafenwaMoses / ImageAI

A python library built to empower developers to build applications and systems with self-contained Computer Vision capabilities
https://www.genxr.co/#products
MIT License
8.5k stars 2.18k forks source link

How to edit yolov3 model(models.py) in order to be able to detect custom objects, classess = 1 #129

Closed dreambit closed 5 years ago

dreambit commented 5 years ago

I have custom image dataset, classess = 1, model is trained on this dataset. i have an errior _Dimension 0 in both shapes must be equal, but are 1 and 18. Shapes are [1,1,1024,255] and [18,1024,1,1]. for 'Assign360' (op: 'Assign') with input shapes: [1,1,1024,255], [18,1024,1,1]

Could you specify how to adjust model to handle this?

OlafenwaMoses commented 5 years ago

ImageAI currently doesn't support training your own custom detection model. It only supports training custom recognition/classification models, and that requires that you have at least 2 classes of objects in your dataset.

For training custom detection models, refer to this conversation. #8