Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Switch to Keras Mish implementation for TfLite compatibility #60

Open
wants to merge 9 commits into
base: master
Choose a base branch
from

Conversation

m-romanenko
Copy link
Collaborator

@m-romanenko m-romanenko commented Jul 6, 2020

Waiting for #61 to be resolved before merging

@@ -0,0 +1,5 @@
"""Activations layers"""

from .mish import Mish
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Mets un path absolu plutot que relatif

>>> X = Mish()(X_input)
"""

def __init__(self, **kwargs):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pas besoin de définir l'init si tu ne fais rien de plus

@click.option(
"--weights_path", default=None, help="Path to .h5 file with model weights"
)
def main(num_classes, weights_path):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Est-ce que tu peux :

  • ne pas appeler ca main()
  • ajouter dans le setup.py une commande comme c'est fait pour la conversion de poids ? Ca permet d'avoir une CLI quand tu installes la librairie

weights_path (str, optional): Path to .h5 pre-trained weights file
"""
model = YOLOv4(
input_shape=(HEIGHT, WIDTH, 3),
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Height and width are parametrizable, is it an argument stored in the tflite model or is it just used for the conversion? We want to make sure users can proceed the inference on any size

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants