What is the output of the hand_detector.tflite
file using the hand landmark detection model hand_landmark.task
locally?
#5616
Labels
platform:python
MediaPipe Python issues
task:hand landmarker
Issues related to hand landmarker: Identify and track hands and fingers
type:support
General questions
type:tflite
TensorFlow Lite
What is the output of the
hand_detector.tflite
file using the hand landmark detection modelhand_landmark.task
locally?I tried to use the
get_output_details
function to output the model's outputs, and the results are as follows:input_details:
output_details:
index: 279
name: Identity
shape: [ 1 2016 18]
dtype: <class 'numpy.float32'>
index: 276
name: Identity_1
shape: [ 1 2016 1]
dtype: <class 'numpy.float32'>
The number 2016, as I understand it, refers to the 2016 SSD anchor boxes within a 192x192 resolution image. The output of 120161 is understood as the score for each anchor box. But what does the output of 1201618 represent? Could it be that the model only selects 2016 pixel points, and the number 18 represents the size information of the anchor boxes of different scales for each pixel point?
The text was updated successfully, but these errors were encountered: