Merge pull request #50 from unarxiv/dev

Merge Object Detection API into codebase

Pipfile

@@ -0,0 +1,46 @@
+[[source]]
+
+name = "pypi"
+url = "https://pypi.python.org/simple"
+verify_ssl = true
+
+
+[dev-packages]
+
+
+
+[packages]
+
+bleach = "==2.1.3"
+certifi = "==2018.4.16"
+chardet = "==3.0.4"
+cycler = "==0.10.0"
+decorator = "==4.3.0"
+"enum34" = "==1.1.6"
+"html5lib" = "==1.0.1"
+idna = "==2.6"
+markdown = "==2.6.11"
+networkx = "==2.1"
+numpy = "==1.14.3"
+pillow = "==5.1.0"
+pkginfo = "==1.4.2"
+protobuf = "==3.5.2"
+pyparsing = "==2.2.0"
+python-dateutil = "==2.7.3"
+pytz = "==2018.4"
+pywavelets = "==0.5.2"
+requests = "==2.18.4"
+requests-toolbelt = "==0.8.0"
+scikit-image = "==0.14.0"
+scipy = "==1.1.0"
+tensorflow-gpu = "==1.8.0"
+tensorflow-tensorboard = "==1.5.1"
+tensorlayer = "==1.8.5"
+tqdm = "==4.23.4"
+twine = "==1.11.0"
+"urllib3" = "==1.23"
+werkzeug = "==0.14.1"
+coverage = "*"
+easydict = "*"
+logbook = "*"
+lxml = "*"

cvtron/3rdparty/.gitignore

@@ -0,0 +1,2 @@
+object_detection/*
+!object_detection/install-ubuntu.sh

cvtron/Pipfile

@@ -16,7 +16,6 @@ python_version = "3.6"
 
 [packages]
 
-tensorflow-gpu = ">=1.4,<1.5"
 tensorlayer = "*"
 tqdm = "*"
 requests = "*"
@@ -25,5 +24,8 @@ easydict = "*"
 yapf = "*"
 autoflake = "*"
 isort = "*"
-logbook = "*"
 cython = "*"
+pycocotools = "*"
+"tensor2tensor" = "<1.4.3"
+logbook = "==1.3.3"
+tensorflow-gpu = "==1.5.0"

cvtron/data_zoo/coco/pycocotools/coco.py

@@ -44,18 +44,21 @@
 # Code written by Piotr Dollar and Tsung-Yi Lin, 2014.
 # Licensed under the Simplified BSD License [see bsd.txt]
 
+import copy
+import itertools
 import json
+import os
+import sys
 import time
+from collections import defaultdict
+
 import matplotlib.pyplot as plt
+import numpy as np
 from matplotlib.collections import PatchCollection
 from matplotlib.patches import Polygon
-import numpy as np
-import copy
-import itertools
+
 from . import mask as maskUtils
-import os
-from collections import defaultdict
-import sys
+
 PYTHON_VERSION = sys.version_info[0]
 if PYTHON_VERSION == 2:
     from urllib import urlretrieve
@@ -430,4 +433,4 @@ def annToMask(self, ann):
         """
         rle = self.annToRLE(ann)
         m = maskUtils.decode(rle)
-        return m
+        return m

cvtron/data_zoo/coco/pycocotools/cocoeval.py

@@ -1,11 +1,14 @@
 __author__ = 'tsungyi'
 
-import numpy as np
+import copy
 import datetime
 import time
 from collections import defaultdict
+
+import numpy as np
+
 from . import mask as maskUtils
-import copy
+
 
 class COCOeval:
     # Interface for evaluating detection on the Microsoft COCO dataset.
@@ -531,4 +534,4 @@ def __init__(self, iouType='segm'):
             raise Exception('iouType not supported')
         self.iouType = iouType
         # useSegm is deprecated
-        self.useSegm = None
+        self.useSegm = None

cvtron/data_zoo/coco/pycocotools/mask.py

@@ -100,4 +100,4 @@ def toBbox(rleObjs):
     if type(rleObjs) == list:
         return _mask.toBbox(rleObjs)
     else:
-        return _mask.toBbox([rleObjs])[0]
+        return _mask.toBbox([rleObjs])[0]

cvtron/data_zoo/coco/setup.py

@@ -1,5 +1,5 @@
-from setuptools import setup, Extension
 import numpy as np
+from setuptools import Extension, setup
 
 # To compile and install locally run "python setup.py build_ext --inplace"
 # To install library to Python site-packages run "python setup.py build_ext install"

cvtron/data_zoo/compress_util.py

@@ -0,0 +1,55 @@
+import os
+import zipfile
+
+from tqdm import tqdm
+
+from cvtron.utils.logger.Logger import logger
+
+
+class ArchiveFile(object):
+    def __init__(self, filepath):
+        if zipfile.is_zipfile(filepath):
+            self._af = zipfile.ZipFile(filepath, 'r')
+            self.filepath = filepath
+            logger.info("Successfully load file" + filepath)
+        else:
+            logger.error("Cannot load file" + filepath)
+
+    def getInfo(self):
+        return self._af.infolist()
+
+    def unzip(self, extractTo, deleteOrigin=False):
+        uncompress_size = sum((file.file_size for file in self._af.infolist()))
+        extracted_size = 0
+        pbar = tqdm(total=uncompress_size, initial=extracted_size, unit='B', unit_scale=True, desc="uncompressing " + self.filepath)
+        for file in self._af.infolist():
+            self._af.extract(file, extractTo)
+            extracted_size += file.file_size
+            pbar.update(extracted_size)
+        self._af.extractall(extractTo)
+        logger.info("Successfully unzip file")
+        if deleteOrigin:
+            os.remove(self.filepath)
+            logger.info("Successfully delete original file")
+
+class ToArchiveFolder(object):
+    def __init__(self, folderPath):
+        if not os.path.exists(folderPath):
+            logger.error("Folder Path Not Found")
+        else:
+            self.folderPath = folderPath
+    def zip(self, target):
+        if self.folderPath is None:
+            logger.error("Folder Path Not Specified")
+            return False
+        else:
+            abs_src = os.path.abspath(self.folderPath)
+            contents = os.walk(self.folderPath)
+            self._af = zipfile.ZipFile(target, 'w', zipfile.ZIP_DEFLATED)
+            for root, dirs, files in contents:
+                for file in files:
+                    abs_name = os.path.abspath(os.path.join(root, file))
+                    arcname = abs_name[len(abs_src) + 1:]
+                    logger.info('adding file' + abs_name)
+                    self._af.write(abs_name, arcname)
+            self._af.close()

cvtron/data_zoo/vlabel/parser.py

@@ -0,0 +1,9 @@
+"""
+Files in this directory is used to handle dataset from Visual Label 
+(https://github.com/unarxiv/visual-label)
+
+The dataset format documentation could be found at 
+(https://cvtron.unarxiv.org/guide/data-format.html)
+"""
+
+from cvtron.data_zoo.compress_util import ArchiveFile

cvtron/model_zoo/constant.py

@@ -1,11 +1,14 @@
 '''
 ORIGINAL URL
+'''
 INCEPTION_CKPT_URL = 'https://deepbase-1251068397.cos.ap-guangzhou.myqcloud.com/inception_v3.ckpt'
 VGG_NPY_URL = 'https://deepbase-1251068397.cos.ap-guangzhou.myqcloud.com/vgg19.npy'
 LAPSRN_URL = 'https://deepbase-1251068397.cos.ap-guangzhou.myqcloud.com/lapsrn.npz'
 YOLO_TINY_CKPT_URL = 'https://deepbase-1251068397.cos.ap-guangzhou.myqcloud.com/yolo_tiny.ckpt'
 '''
+CDN URL
 INCEPTION_CKPT_URL = 'http://cdn.zhitantech.com/static/inception_v3.ckpt'
 VGG_NPY_URL = 'http://cdn.zhitantech.com/static/vgg19.npy'
 LAPSRN_URL = 'http://cdn.zhitantech.com/static/lapsrn.npz'
 YOLO_TINY_CKPT_URL = 'http://cdn.zhitantech.com/static/yolo_tiny.ckpt'
+'''

cvtron/modeling/detector/api.py

@@ -5,6 +5,15 @@
 from cvtron.utils.config_loader import MODEL_ZOO_PATH
 from cvtron.utils.reporter import print_detect_result
 
+# from object_detection.object_detection_class import ObjectDetection
+
+train_config = {'pipeline_config_file':'/home/wujia/examples/platform/test-platform/CVTron/cvtron/object_detection/samples/configs/ssd_inception_v2_pets.config',
+          'train_dir':'/home/wujia/examples/platform/test-platform/CVTron/cvtron/tests/train_dir',
+          'gpu_id':1,
+          'log_every_n_steps':100}
+
+infer_config = {'model_name':'yolo'}
+
 
 def simple_detect_api(img_file,
                       model_name='yolo_tiny',
@@ -23,3 +32,13 @@ def get_detector(model_name='yolo_tiny',
         raise NotImplementedError
     objectDetector = ObjectDetector(model_name, model_path)
     return objectDetector
+
+def get_object_detector(configs):
+    # detector = ObjectDetection(configs)
+    return None
+
+def get_train_config():
+    return train_config
+
+def get_infer_config():
+    return infer_config

cvtron/modeling/detector/slim_object_detector.py

@@ -0,0 +1,115 @@
+#coding:utf-8
+import numpy as np
+from PIL import Image
+import tensorflow as tf
+from object_detection.utils import label_map_util
+
+from cvtron.Base.decorator import singleton
+from cvtron.utils.logger.Logger import logger
+from cvtron.utils.image_loader import get_image_size
+from cvtron.utils.image_loader import load_image_into_numpy_array
+
+@singleton
+class SlimObjectDetector(object):
+    def __init__(self):
+        self.hasInitialized = False
+
+
+    def set_label_map(self, label_map_file):
+        self.label_map = label_map_file
+
+    def init(self, model_path):
+        if not self.hasInitialized:
+            self._init_model_(model_path)
+        else:
+            logger.info('model has initialized, skipping')
+
+    def _init_model_(self, model_path):
+        logger.info('initiating model')
+        detection_graph = tf.Graph()
+        with detection_graph.as_default():
+            od_graph_def = tf.GraphDef()
+            with tf.gfile.GFile(model_path, 'rb') as fid:
+                serialized_graph = fid.read()
+                od_graph_def.ParseFromString(serialized_graph)
+                tf.import_graph_def(od_graph_def, name='')
+        self.graph = detection_graph
+        logger.info('model initialized')
+
+    def format_output(self, results, im_width, im_height, threshold):
+        label_map = label_map_util.load_labelmap(self.label_map)
+        categories = label_map_util.convert_label_map_to_categories(label_map, max_num_classes=90, use_display_name=True)
+        category_index = label_map_util.create_category_index(categories)
+        vis_results = []
+        for i in range(len(results)):
+            output_dict = results[i]
+            boxes = output_dict['detection_boxes']
+            classes = output_dict['detection_classes']
+            scores = output_dict['detection_scores']
+            for j in range(boxes.shape[0]):
+                if scores[i] > threshold:
+                    result = {}
+                    ymin, xmin, ymax, xmax = tuple(boxes[i].tolist())
+                    result['x_min'] = xmin * im_width
+                    result['x_max'] = xmax * im_width
+                    result['y_min'] = ymin * im_height
+                    result['y_max'] = ymax * im_height
+                    result['score'] = float(scores[i])
+                    if classes[i] in category_index.keys():
+                        class_name = category_index[classes[i]]['name']
+                    else:
+                        class_name = 'N/A'
+                    result['class_name'] = str(class_name)
+                    if result not in vis_results:
+                        vis_results.append(result)
+        return vis_results
+
+    def detect(self, img_file, threshold = 0.7):
+        # read image
+        image_np = load_image_into_numpy_array(img_file)
+        (im_width, im_height) = get_image_size(img_file)
+        # Inference process
+        with self.graph.as_default():
+            with tf.Session() as sess:
+                ## Get handles to input and output tensors
+                ops = tf.get_default_graph().get_operations()
+                all_tensor_names = {
+                    output.name for op in ops for output in op.outputs
+                }
+                tensor_dict = {}
+                for key in [
+                    'num_detections', 'detection_boxes', 'detection_scores',
+                    'detection_classes', 'detection_masks'
+                ]:
+                    tensor_name = key + ':0'
+                    if tensor_name in all_tensor_names:
+                        tensor_dict[key] = tf.get_default_graph().get_tensor_by_name(tensor_name)
+                if 'detection_masks' in tensor_dict:
+                    # The following processing is only for single image
+                    detection_boxes = tf.squeeze(tensor_dict['detection_boxes'], [0])
+                    detection_masks = tf.squeeze(tensor_dict['detection_masks'], [0])
+                    # Reframe is required to translate mask from box coordinates to image coordinates and fit the image size.
+                    real_num_detection = tf.cast(tensor_dict['num_detections'][0], tf.int32)
+                    detection_boxes = tf.slice(detection_boxes, [0, 0], [real_num_detection, -1])
+                    detection_masks = tf.slice(detection_masks, [0, 0, 0], [real_num_detection, -1, -1])
+                    detection_masks_reframed = ops.reframe_box_masks_to_image_masks(
+                    detection_masks, detection_boxes, image_np.shape[0], image_np.shape[1])
+                    detection_masks_reframed = tf.cast(
+                        tf.greater(detection_masks_reframed, 0.5), tf.uint8)
+                    # Follow the convention by adding back the batch dimension
+                    tensor_dict['detection_masks'] = tf.expand_dims(
+                        detection_masks_reframed, 0)
+                image_tensor = tf.get_default_graph().get_tensor_by_name('image_tensor:0')
+                ## Run inference
+                output_dict = sess.run(tensor_dict, feed_dict={
+                    image_tensor: np.expand_dims(image_np, 0)
+                })
+                output_dict['num_detections'] = int(output_dict['num_detections'][0])
+                output_dict['detection_classes'] = output_dict[
+                    'detection_classes'][0].astype(np.uint8)
+                output_dict['detection_boxes'] = output_dict['detection_boxes'][0]
+                output_dict['detection_scores'] = output_dict['detection_scores'][0]
+                if 'detection_masks' in output_dict:
+                    output_dict['detection_masks'] = output_dict['detection_masks'][0]
+                results = [output_dict]
+                return self.format_output(results, im_width, im_height, threshold)

cvtron/modeling/segmentor/ImageSegmentor.py

@@ -21,7 +21,6 @@ def __init__(self, model_path, args):
         self.network = deeplab_v3(self.x, args, is_training=False, reuse=False)
         self.pred = tf.argmax(self.network, axis=3)
     def _init_model_(self):
-        print('init model')
         saver = tf.train.Saver()
         saver.restore(self.sess, os.path.join(self.model_path,'deeplabv3/model.ckpt'))
     def segment(self, img_file):

cvtron/modeling/segmentor/api.py

@@ -37,5 +37,4 @@ def get_segmentor_trainer(config):
     return dlt
 
 def get_defaultConfig():
-    print(config)
     return config

cvtron/trainers/base/BaseTrainer.py

@@ -0,0 +1,5 @@
+class BaseTrainer(object):
+    def __init__(self):
+        pass
+    def start(self):
+        pass