TensorFlow - how to detect multi-objects in one image


How to detect multi-objects( different or same type) with machine learning.
I have tested successfully for one detection with this command
tf.classify(net, img, min_scale=1.0, scale_mul=0.8, x_overlap=0.8, y_overlap=0.8)

Please advise how to do the multi -detection.

Hi, we don’t have a SSD type network yet to detect multiple objects at the same time. Right now it’s just image classification.

We are working with Edge Impulse to release:

  1. A segmentation network architecture to support binary segmentation images.
  2. A multi-bounding box detection architecture.

Until these networks exist we cannot add the feature to the camera. However, we expect they will exist in the future soon.

Until then, you just have a sliding window detector.