snapshot...

modules/deeplearning/cloud_opd_srcnn_viirs.py

@@ -693,6 +693,7 @@ def run_restore_static(directory, ckpt_dir, out_file=None):
 
 
 def run_evaluate_static(in_file, out_file, ckpt_dir):
+    border = int((KERNEL_SIZE - 1) / 2)
 
     h5f = h5py.File(in_file, 'r')
 
@@ -705,6 +706,10 @@ def run_evaluate_static(in_file, out_file, ckpt_dir):
     ylen, xlen = bt.shape
     bt = bt[int(ylen/2):ylen, (int(xlen/2)-400):(int(xlen/2)+400)]
 
+    cld_opd_orig = get_grid_values_all(h5f, 'orig/' + label_param)
+    ylen, xlen = cld_opd_orig.shape
+    cld_opd_orig = cld_opd_orig[int(ylen/2):ylen, (int(xlen/2)-400):(int(xlen/2)+400)]
+
     cld_opd = get_grid_values_all(h5f, 'super/' + label_param)
     ylen, xlen = cld_opd.shape
     cld_opd = cld_opd[int(ylen/2):ylen, (int(xlen/2)-800):(int(xlen/2)+800)]
@@ -743,6 +748,13 @@ def run_evaluate_static(in_file, out_file, ckpt_dir):
     cld_opd = smooth_2d(cld_opd)
     cld_opd = normalize(cld_opd, label_param, mean_std_dct)
 
+    # cld_opd = np.where(np.isnan(cld_opd_orig), 0, cld_opd_orig)
+    # cld_opd = cld_opd[nn.slc_y_m, nn.slc_x_m]
+    # cld_opd = np.expand_dims(cld_opd, axis=0)
+    # cld_opd = nn.upsample(cld_opd)
+    # cld_opd = smooth_2d(cld_opd)
+    # cld_opd = normalize(cld_opd, label_param, mean_std_dct)
+
     data = np.stack([bt, refl, cld_opd], axis=3)
 
     h5f.close()
@@ -752,14 +764,19 @@ def run_evaluate_static(in_file, out_file, ckpt_dir):
     # cld_opd_sres = descale(cld_opd_sres, label_param, mean_std_dct)
     _, ylen, xlen, _ = cld_opd_sres.shape
 
-    cld_opd_sres_out = np.zeros((LEN_Y, LEN_X), dtype=np.float32)
-    refl_out = np.zeros((LEN_Y, LEN_X), dtype=np.float32)
-    cld_opd_out = np.zeros((LEN_Y, LEN_X), dtype=np.float32)
-
-    border = int((KERNEL_SIZE - 1) / 2)
-    cld_opd_sres_out[border:(border+ylen), border:(border+xlen)] = cld_opd_sres[0, :, :, 0]
-    refl_out[0:(ylen+2*border), 0:(xlen+2*border)] = refl[0, :, :]
-    cld_opd_out[0:(ylen+2*border), 0:(xlen+2*border)] = cld_opd[0, :, :]
+    # cld_opd_sres_out = np.zeros((LEN_Y, LEN_X), dtype=np.float32)
+    # refl_out = np.zeros((LEN_Y, LEN_X), dtype=np.float32)
+    # cld_opd_out = np.zeros((LEN_Y, LEN_X), dtype=np.float32)
+    #
+    # cld_opd_sres_out[border:(border+ylen), border:(border+xlen)] = cld_opd_sres[0, :, :, 0]
+    # refl_out[0:(ylen+2*border), 0:(xlen+2*border)] = refl[0, :, :]
+    # cld_opd_out[0:(ylen+2*border), 0:(xlen+2*border)] = cld_opd[0, :, :]
+
+    cld_opd_sres_out = cld_opd_sres[0, :, :, 0]
+    refl_out = refl[0, border:(ylen-border), border:(xlen-border)]
+    cld_opd_out = cld_opd[0, border:(ylen-border), border:(xlen-border)]
+    cld_opd_hres = cld_opd_hres[border:(ylen-border), border:(xlen-border)]
+    print(cld_opd_sres_out.shape, refl_out.shape, cld_opd_out.shape, cld_opd_hres.shape)
 
     refl_out = denormalize(refl_out, 'refl_0_65um_nom', mean_std_dct)
     cld_opd_out = denormalize(cld_opd_out, label_param, mean_std_dct)