Downloading ImageAI in Raspberry Pi?

I want to download the model chosen after training in Raspberry Pi. I found instructions on how to download dependencies such as OpenCV, Tensorflow and Keras but I am unsure of the steps necessary to download ImageAI into the Raspberry Pi. Please can someone guide me on how to download ImageAI library and the model I chose so that I can run the model to detect objects real-time?

1 Like

You can clone the repository and then run the setup.py

git clone https://github.com/OlafenwaMoses/ImageAI.git

then

cd ImageAI

and
python3 setup.py install

@SruthiRam Also note that you can’t run the detection in real-time on the Pi. The maximum you can get is a single image frame per second.

Oh okay I didn’t know that. Thanks for your help. Is there anyway to create checkpoints to save the number of epochs created while training the models? Google Colab stops after about 12 hours and only about half of the epochs are over by that time.I have to start all the way from the beginning again. I tried using the ModelCheckpoint from Keras but I got an Name error stating that ModelCheckpoint was not defined even after installing it in the Google Colab notebook. Please can you help regarding this matter.

1 Like

@SruthiRam You can mount your Google Drive for the training and place your dataset in a Google drive folder so that all the models generated during training will be saved in the that folder. Then you can use the model with the minimal loss as the pre-trained model for the next training.

1 Like

Ohh okay thank you very much!