Commit 54710be6 authored by 刘丹's avatar 刘丹
Browse files

update

parent 1f2219cd

eval_all_ckpt_dice_multi.py

+2 −37
Original line number Diff line number Diff line
@@ -16,7 +16,7 @@ tf.app.flags.DEFINE_string('heatmap_output_dir', '../result/oxford-test-result/h 
tf.app.flags.DEFINE_bool('no_write_images', False, 'do not write images')



import resnet_v1_model_dice_multi as model

from oxford_R01 import restore_rectangle

from oxford import restore_rectangle



FLAGS = tf.app.flags.FLAGS

CKPT_PATH = None#'../model/hand_resnet_v1_50_dice_multi_oxford_aug_rbox/model.ckpt-98051'#
@@ -32,12 +32,6 @@ def get_images(): 
    '''

    files = []

    exts = ['jpg', 'png', 'jpeg', 'JPG']

    # for parent, dirnames, filenames in os.walk(FLAGS.test_data_path):

    #     for filename in filenames:

    #         for ext in exts:

    #             if filename.endswith(ext):

    #                 files.append(os.path.join(parent, filename))

    #                 break

    for ext in exts:

        files += glob.glob(os.path.join(FLAGS.test_data_path, '*.' + ext))

    print('Find {} images'.format(len(files)))
@@ -103,35 +97,6 @@ def detect(im_fn, im, ratio_h, ratio_w, score_map, geo_map, timer, score_map_thr 
    # updated

    # resize_score_map = cv2.resize(score_map, (score_map.shape[1]*4, score_map.shape[0]*4))

    # cv2.imwrite(FLAGS.heatmap_output_dir+os.path.basename(im_fn)[:-4]+'_heatmap.png', resize_score_map*255)

    # #generate the heatmap

    # xy_text_0 = np.argwhere(score_map >= 0)

    # # sort the text boxes via the y axis

    # xy_text_0 = xy_text_0[np.argsort(xy_text_0[:, 0])]

    # # restore

    # text_box_restored_0 = restore_rectangle(xy_text_0[:, ::-1]*4, geo_map[xy_text_0[:, 0], xy_text_0[:, 1], :])

    # boxes_0 = np.zeros((text_box_restored_0.shape[0], 9), dtype=np.float32)

    # boxes_0[:, :8] = np.clip(text_box_restored_0.reshape((-1, 8)), 0, max(score_map.shape[0]*4, score_map.shape[1]*4))

    # boxes_0[:, 8] = score_map[xy_text_0[:, 0], xy_text_0[:, 1]]

    # # mask_0 = np.zeros((score_map.shape[0]*4, score_map.shape[1]*4), dtype=np.float32)

    # # overlap_count = np.zeros((score_map.shape[0]*4, score_map.shape[1]*4))

    # heatmap = {}

    # heatmap["image_name"] = os.path.basename(im_fn)

    # bboxes = []

    # if boxes_0 is not None:

    #     for box_0 in boxes_0:

    #         if round(box_0[8], 6) > 0:

    #             pos = box_0[:8].reshape((4, 2))

    #             pos[:, 0] /= ratio_w

    #             pos[:, 1] /= ratio_h

    #             pos = sort_poly(pos.astype(np.int32))

    #             # print pos[0, 1], pos[2, 1], pos[0, 0], pos[2, 0]

    #             bboxes.append({"bbox":list([int(pos[0, 0]), int(pos[0, 1]), int(pos[2, 0]), int(pos[2, 1])]), "score":round(box_0[8], 6)})

    #             # mask_0[pos[0, 1]:pos[2, 1], pos[0, 0]:pos[2, 0]] += box_0[8]

    #             # overlap_count[pos[0, 1]:pos[2, 1], pos[0, 0]:pos[2, 0]] += 1

    # heatmap["bboxes"] = bboxes

    # heatmaps.append(heatmap)

    # final_heat_map = np.true_divide(mask_0, overlap_count)

    # cv2.imwrite(FLAGS.heatmap_output_dir+os.path.basename(im_fn)[:-4]+'_heatmap.png', final_heat_map*255)

    

    # filter the score map

    xy_text = np.argwhere(score_map > score_map_thresh)

multigpu_train_dice_multi.py

+2 −2
Original line number Diff line number Diff line
@@ -16,8 +16,8 @@ tf.app.flags.DEFINE_integer('save_checkpoint_steps', 2000, '') 
tf.app.flags.DEFINE_integer('save_summary_steps', 2000, '')

tf.app.flags.DEFINE_string('pretrained_model_path', '../model/resnet50/resnet_v1_50.ckpt', '')



import resnet_v1_model_dice_multi_alpha_p5 as model

import oxford_R01 as data_processor

import resnet_v1_model_dice_multi as model

import oxford as data_processor



FLAGS = tf.app.flags.FLAGS

multigpu_train_dice_multi_weighted_fusion.py

+2 −2
Original line number Diff line number Diff line
@@ -16,8 +16,8 @@ tf.app.flags.DEFINE_integer('save_checkpoint_steps', 2000, '') 
tf.app.flags.DEFINE_integer('save_summary_steps', 2000, '')

tf.app.flags.DEFINE_string('pretrained_model_path', '../model/resnet50/resnet_v1_50.ckpt', '')



import resnet_v1_model_dice_multi_beta_1_weighted_fusion as model

import oxford_R01 as data_processor

import resnet_v1_model_dice_multi_weighted_fusion as model

import oxford as data_processor



FLAGS = tf.app.flags.FLAGS

multigpu_train_vgg16_dice_multi.py

+1 −1
Original line number Diff line number Diff line
@@ -18,7 +18,7 @@ tf.app.flags.DEFINE_string('pretrained_model_path', '../model/vgg16/vgg_16.ckpt' 
# tf.app.flags.DEFINE_string('pretrained_model_path', None, '')



import vgg16_model_dice_multi as model

import oxford_R01 as data_processor

import oxford as data_processor



FLAGS = tf.app.flags.FLAGS

multigpu_train_vgg16_dice_multi_weighted_fusion.py

+1 −1
Original line number Diff line number Diff line
@@ -18,7 +18,7 @@ tf.app.flags.DEFINE_string('pretrained_model_path', '../model/vgg16/vgg_16.ckpt' 
# tf.app.flags.DEFINE_string('pretrained_model_path', None, '')



import vgg16_model_dice_multi_weighted_fusion as model

import oxford_R01 as data_processor

import oxford as data_processor



FLAGS = tf.app.flags.FLAGS