Skip to content

Instantly share code, notes, and snippets.

@meetps
Created March 2, 2017 17:35
Show Gist options
  • Save meetps/4a5e61aa0dd4580f76e95ea0afff497d to your computer and use it in GitHub Desktop.
Save meetps/4a5e61aa0dd4580f76e95ea0afff497d to your computer and use it in GitHub Desktop.
# def predict_id(id, model, trs):
# img = utils.M(id)
# x = utils.stretch_n(img)
# cnv = np.zeros((960, 960, 8)).astype(np.float32)
# prd = np.zeros((n_classes, 960, 960)).astype(np.float32)
# cnv[:img.shape[0], :img.shape[1], :] = x
# for i in range(0, 6):
# line = []
# for j in range(0, 6):
# line.append(cnv[i * ISZ:(i + 1) * ISZ, j * ISZ:(j + 1) * ISZ])
# x = 2 * np.transpose(line, (0, 3, 1, 2)) - 1
# x = np.transpose(x, (0,2,3,1))
# tmp = model.predict(x, batch_size=4)
# for j in range(tmp.shape[0]):
# prd[:, i * ISZ:(i + 1) * ISZ, j * ISZ:(j + 1) * ISZ] = tmp[j]
# # trs = [0.4, 0.1, 0.4, 0.3, 0.3, 0.5, 0.3, 0.6, 0.1, 0.1]
# for i in range(n_classes):
# prd[i] = prd[i] > trs[i]
# return prd[:, :img.shape[0], :img.shape[1]]
# def check_predict(id='6120_2_3'):
# params = {'input_shape':(n_channels, img_rows, img_cols),
# 'n_classes':n_classes,
# 'feat_scale':feat_scale}
# model = unet(params)
# model.load_weights('./savedModel/myModel.hdf5')
# model.compile(optimizer=Adam(),
# loss='binary_crossentropy',
# metrics=[jaccard_coef, jaccard_coef_int, 'accuracy'])
# msk = predict_id(id, model, [0.4, 0.1, 0.4, 0.3, 0.3, 0.5, 0.3, 0.6, 0.1, 0.1])
# img = M(id)
# plt.figure()
# ax1 = plt.subplot(121)
# ax1.set_title('image ID:6120_2_3')
# ax1.imshow(img[:, :, 5], cmap=plt.get_cmap('gist_ncar'))
# ax2 = plt.subplot(122)
# ax2.set_title('predict bldg pixels')
# ax2.imshow(msk[0], cmap=plt.get_cmap('gray'))
# plt.show()
# ax3 = plt.subplot(133)
# ax3.set_title('predict bldg polygones')
# ax3.imshow(mask_for_polygons(mask_to_polygons(msk[0], epsilon=1), img.shape[:2]), cmap=plt.get_cmap('gray'))
############ test
# params = {'input_shape':(n_channels, img_rows, img_cols),
# 'n_classes':n_classes,
# 'feat_scale':feat_scale}
# model = unet(params)
# model.load_weights('./savedModel/myModel.hdf5')
# model.compile(optimizer=Adam(),
# loss='binary_crossentropy',
# metrics=[jaccard_coef, jaccard_coef_int, 'accuracy'])
# id='6120_2_3'
# trs = [0.4, 0.1, 0.4, 0.3, 0.3, 0.5, 0.3, 0.6, 0.1, 0.1]
# img = np.load('results/img.npy')
# x = np.load('results/x.npy')
# cnv = np.zeros((960, 960, 8)).astype(np.float32)
# prd = np.zeros((n_classes, 960, 960)).astype(np.float32)
# cnv[:img.shape[0], :img.shape[1], :] = x
# for i in range(0, 6):
# line = []
# for j in range(0, 6):
# line.append(cnv[i * ISZ:(i + 1) * ISZ, j * ISZ:(j + 1) * ISZ])
# x = 2 * np.transpose(line, (0, 3, 1, 2)) - 1
# x = np.transpose(x, (0,2,3,1))
# tmp = model.predict(x, batch_size=4)
# for j in range(tmp.shape[0]):
# prd[:, i * ISZ:(i + 1) * ISZ, j * ISZ:(j + 1) * ISZ] = tmp[j]
# # trs = [0.4, 0.1, 0.4, 0.3, 0.3, 0.5, 0.3, 0.6, 0.1, 0.1]
# for i in range(n_classes):
# prd[i] = prd[i] > trs[i]
# msk = prd[:, :img.shape[0], :img.shape[1]]
# # img = M(id)
# plt.figure()
# ax1 = plt.subplot(121)
# ax1.set_title('image ID:6120_2_3')
# ax1.imshow(img[:, :, 5], cmap=plt.get_cmap('gist_ncar'))
# ax2 = plt.subplot(122)
# ax2.set_title('predict bldg pixels')
# ax2.imshow(msk[0], cmap=plt.get_cmap('gray'))
# plt.show()
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment