burke86 / astro_rcnn Goto Github PK

Hello!
How to download weights?
$ git lfs fetch .
fetch: Fetching reference refs/heads/master
batch response: This repository is over its data quota. Account responsible for LFS bandwidth should purchase more data packs to restore access.
error: failed to fetch some objects from 'https://github.com/burke86/astro_rcnn.git/info/lfs'

Requires generalizing load_image_disk at the very least. May need to generalize Mask R-CNN code to handle > 3 image channels.

See https://github.com/facebookresearch/detectron2/tree/master/projects/TensorMask

We may want to start a new repo for this to interface with detectron2.

hello,

I try to run this "https://nbviewer.org/github/burke86/astrodet/blob/main/demo_decam.ipynb ". . in this code while i run

nsample = 3 # Number of example images to plot
fig, axs = plt.subplots(1, nsample, figsize=(5*nsample, 5))

for i, d in enumerate(random.sample(dataset_dicts['test'], nsample)):

an error comes " ValueError: Sample larger than population or is negative"

so how to fix it?

Hi all,

Maybe it is a dumb question, but it is my first time playing with deep learning codes.

To use this code, should I train the neural network on my images? My final goal would be to create, given an image in a given band, a mask to mask sources out (galaxies and stars) in order to deal with the background only.

Thanks in advance

Currently the masks are roughly defined using a threshold enclosing pixel values normalized to the maximum value of the source. Then, some smoothing is done to the mask. We should investigate using a Petrosian ellipse for a more standardized mask.

The statmorph module has codes to compute it, see '_segmap_gini' code: https://statmorph.readthedocs.io/en/latest/_modules/statmorph/statmorph.html

We may want to implement a similar version of this function.

The AP score is not a probably not the most sensitive metric for testing deblending. Should try others, for example the purity metric using a ratio of weighted sums: rho_i = sum G_i G_i / Sum G_i (G_1 + G_2 + ...) from https://project.lsst.org/meetings/lsst2018/sites/lsst.org.meetings.lsst2018/files/Blending%20Session%205%20%28Tools%20and%20Data%20Sets%29%20-%20Slides_0.pdf#page=13&zoom=100,0,0