Merge branch 'main' of https://huggingface.co/datasets/AstroCompress/GBI-16-4D into main
Browse files- utils/eval_baselines.py +9 -9
- utils/sdss_downloading.txt +2 -0
utils/eval_baselines.py
CHANGED
@@ -89,30 +89,30 @@ def main(dim):
|
|
89 |
|
90 |
for path in tqdm(file_paths):
|
91 |
with fits.open(path) as hdul:
|
92 |
-
if dim == '2d':
|
93 |
arr = hdul[0].data[0][2]
|
94 |
arrs = [arr]
|
95 |
-
elif dim == '2d-top':
|
96 |
arr = hdul[0].data[0][2]
|
97 |
arr = split_uint16_to_uint8(arr)[0]
|
98 |
arrs = [arr]
|
99 |
-
elif dim == '2d-bottom':
|
100 |
arr = hdul[0].data[0][2]
|
101 |
arr = split_uint16_to_uint8(arr)[1]
|
102 |
arrs = [arr]
|
103 |
-
elif dim == '3dt' and len(hdul[0].data) > 2:
|
104 |
arr = hdul[0].data[0:3][2]
|
105 |
arrs = [arr]
|
106 |
-
elif dim == '3dw' and len(hdul[0].data[0]) > 2:
|
107 |
arr = hdul[0].data[0][0:3]
|
108 |
arrs = [arr]
|
109 |
-
elif dim == '3dt_reshape' and len(hdul[0].data) > 2:
|
110 |
arr = hdul[0].data[0:3][2].reshape((800, -1))
|
111 |
arrs = [arr]
|
112 |
-
elif dim == '3dw_reshape' and len(hdul[0].data[0]) > 2:
|
113 |
arr = hdul[0].data[0][0:3].reshape((800, -1))
|
114 |
arrs = [arr]
|
115 |
-
elif dim == 'tw':
|
116 |
init_arr = hdul[0].data
|
117 |
def arrs_gen():
|
118 |
for i in range(init_arr.shape[-2]):
|
@@ -153,7 +153,7 @@ if __name__ == "__main__":
|
|
153 |
parser.add_argument(
|
154 |
"dimension",
|
155 |
choices=['2d', '2d-top', '2d-bottom', '3dt', '3dw', 'tw', '3dt_reshape', '3dw_reshape'],
|
156 |
-
help="Specify whether the data is 2d, 3dt (3d time dimension),
|
157 |
)
|
158 |
args = parser.parse_args()
|
159 |
dim = args.dimension.lower()
|
|
|
89 |
|
90 |
for path in tqdm(file_paths):
|
91 |
with fits.open(path) as hdul:
|
92 |
+
if dim == '2d': # compress the first timestep frame, R wavelength band (index 2)
|
93 |
arr = hdul[0].data[0][2]
|
94 |
arrs = [arr]
|
95 |
+
elif dim == '2d-top': # same as 2d, but only top 8 bits. This is to compare with similarly preprocessed neural approaches.
|
96 |
arr = hdul[0].data[0][2]
|
97 |
arr = split_uint16_to_uint8(arr)[0]
|
98 |
arrs = [arr]
|
99 |
+
elif dim == '2d-bottom': # same as 2d, but only bottom 8 bits. This is to compare with similarly preprocessed neural approaches.
|
100 |
arr = hdul[0].data[0][2]
|
101 |
arr = split_uint16_to_uint8(arr)[1]
|
102 |
arrs = [arr]
|
103 |
+
elif dim == '3dt' and len(hdul[0].data) > 2: # 3D tensor with first 3 timestep frames of wavelength band index 2
|
104 |
arr = hdul[0].data[0:3][2]
|
105 |
arrs = [arr]
|
106 |
+
elif dim == '3dw' and len(hdul[0].data[0]) > 2: # 3D tensor with first timestep frame on wavelength bands of indices 1,2,3 (G, R, I bands)
|
107 |
arr = hdul[0].data[0][0:3]
|
108 |
arrs = [arr]
|
109 |
+
elif dim == '3dt_reshape' and len(hdul[0].data) > 2: # Same as 3dt but reshape into 2D array, for compatibility with JPEG-LS and RICE
|
110 |
arr = hdul[0].data[0:3][2].reshape((800, -1))
|
111 |
arrs = [arr]
|
112 |
+
elif dim == '3dw_reshape' and len(hdul[0].data[0]) > 2: # Same as 3dw but reshape into 2D array, for compatibility with JPEG-LS and RICE
|
113 |
arr = hdul[0].data[0][0:3].reshape((800, -1))
|
114 |
arrs = [arr]
|
115 |
+
elif dim == 'tw': # Iterate through all possible arrays where the x,y spatial location is fixed, and the remaining 2D array consists of ALL timesteps, ALL wavelengths.
|
116 |
init_arr = hdul[0].data
|
117 |
def arrs_gen():
|
118 |
for i in range(init_arr.shape[-2]):
|
|
|
153 |
parser.add_argument(
|
154 |
"dimension",
|
155 |
choices=['2d', '2d-top', '2d-bottom', '3dt', '3dw', 'tw', '3dt_reshape', '3dw_reshape'],
|
156 |
+
help="Specify whether the data is 2d, 3dt (3d time dimension), 3dw (3d wavelength dimension), 2d-top (only top 8 bits), 2d-bottom (only bottom 8 bits), tw (only a single x,y spatial location but all timesteps and wavelengths), 3dt_reshape or 3dw_reshape for the 2D flattened 3D evals, for use on JPEG-LS or RICE."
|
157 |
)
|
158 |
args = parser.parse_args()
|
159 |
dim = args.dimension.lower()
|
utils/sdss_downloading.txt
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
Please visit this repo:
|
2 |
+
https://github.com/profjsb/astrocompress/tree/main
|