custom object detection on Google colab & android deployment

This is a rapid prototyping course which will help you to create a wonderful TensorFlow Lite object detection android app within 3 hours!. The student will not require any high-end computer for this course. The only use of our local PC is to do the labelling, generating TensorFlow record files, label_map and android app debugging.

The high-end deep learning training part will be done on Google colaboratory. There will be step by step explanation on each topic covering only the important part of our modeling. We will train our model using the transfer learning method, using TensorFlow Lite specific SSD_mobilenet_v2_quantized architecture. The whole course can be split into 3 parts. The first one is labelling, the second one is training and the third one is android debugging. Labelling takes around 1 hour, training 1.5 hours and debugging 0.5 hours. A rapid prototyping course without compromising the beauty of computer vision.