Scribd
Upload
2 views

Start - Copy (2)

The document outlines a script for setting up and using a TensorFlow object detection model, specifically 'my_ssd_mobnet'. It includes paths for various resources, configuration settings for the model, and code for real-time detection using OpenCV. The script also handles loading the model, processing images, and displaying detection results in a video feed.

Uploaded by

jamesrojjer0012
Copyright
© © All Rights Reserved
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
2 views

Start - Copy (2)

The document outlines a script for setting up and using a TensorFlow object detection model, specifically 'my_ssd_mobnet'. It includes paths for various resources, configuration settings for the model, and code for real-time detection using OpenCV. The script also handles loading the model, processing images, and displaying detection results in a video feed.

Uploaded by

jamesrojjer0012
Copyright
© © All Rights Reserved
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 3

SCRIPTS_PATH = 'Tensorflow/scripts' # generated tf records

APIMODEL_PATH = 'Tensorflow/models' # tf object detection model from github


WORKSPACE_PATH = 'Tensorflow/workspace'
ANNOTATION_PATH = WORKSPACE_PATH + '/annotations'
IMAGE_PATH = WORKSPACE_PATH + '/images'
PRETRAINED_MODEL_PATH = WORKSPACE_PATH + '/pre-trained-models'
MODEL_PATH = WORKSPACE_PATH + '/models' # will store trained models
CONFIG_PATH = MODEL_PATH + '/my_ssd_mobnet/pipeline.config'
CHECKPOINT_PATH = MODEL_PATH + '/my_ssd_mobnet/' # stores checkpoints from trained
models

labels = [
{'name':'Hello', 'id':1},
{'name':'Friend', 'id':2},
{'name':'Yes', 'id':3},
{'name':'No', 'id':4},
{'name':'Thank You', 'id':5},
{'name':'Ok', 'id':6},
{'name':'Bathroom', 'id':7},
{'name':'Please', 'id':8},
{'name':'1000 Years of Death', 'id':9},
]

CUSTOM_MODEL_NAME = 'my_ssd_mobnet' # name of custom model

# Importing dependencies for updating config

import tensorflow as tf
from object_detection.utils import config_util # for path configuration
from object_detection.protos import pipeline_pb2
from google.protobuf import text_format

# Loading trained model from checkpoint


import os
from object_detection.utils import label_map_util
from object_detection.utils import visualization_utils as vz_utils # for making
bounding box
from object_detection.builders import model_builder # builds model from checkpoint

# Real-Time Detections using OpenCV


# importing dependencies
import cv2
import numpy as np

CONFIG_PATH = MODEL_PATH + '/' + CUSTOM_MODEL_NAME + '/pipeline.config'


config = config_util.get_configs_from_pipeline_file(CONFIG_PATH) # getting config
into notebook for making changes

pipeline_config = pipeline_pb2.TrainEvalPipelineConfig() # Template pipeline config


with tf.io.gfile.GFile(CONFIG_PATH, 'r') as f:
proto_str = f.read()
text_format.Merge(proto_str, pipeline_config)

pipeline_config.model.ssd.num_classes = 9 # number of different types of model need


to be configured i.e. 9
pipeline_config.train_config.batch_size = 4 # data processed dependent on hardware
capacity
pipeline_config.train_config.fine_tune_checkpoint = PRETRAINED_MODEL_PATH +
'/ssd_mobilenet_v2_fpnlite_320x320_coco17_tpu-8/checkpoint/ckpt-0'
pipeline_config.train_config.fine_tune_checkpoint_type = 'detection' # Changing
from classification to detection
pipeline_config.train_input_reader.label_map_path = ANNOTATION_PATH +
'/label_map.pbtxt'
pipeline_config.train_input_reader.tf_record_input_reader.input_path[:] =
[ANNOTATION_PATH + '/train.record'] # tf records
pipeline_config.eval_input_reader[0].label_map_path = ANNOTATION_PATH +
'/label_map.pbtxt'
pipeline_config.eval_input_reader[0].tf_record_input_reader.input_path[:] =
[ANNOTATION_PATH + '/test.record'] # tf records

# Load pipeline config and build a detection model


configs = config_util.get_configs_from_pipeline_file(CONFIG_PATH)
detection_model = model_builder.build(model_config=configs['model'],
is_training=False)

# Restore checkpoint
ckpt = tf.compat.v2.train.Checkpoint(model=detection_model)
ckpt.restore(os.path.join(CHECKPOINT_PATH, 'ckpt-21')).expect_partial() # ckpt-21
is the latest checkpoint

@tf.function
def detect_fn(image):
image, shapes = detection_model.preprocess(image) # resizing image to 320x320
prediction_dict = detection_model.predict(image, shapes) # making predictions
detections = detection_model.postprocess(prediction_dict, shapes)
return detections

category_index = label_map_util.create_category_index_from_labelmap(ANNOTATION_PATH
+ '/label_map.pbtxt')

# Setting up capture
cap = cv2.VideoCapture(0) # '0' is the device number for my camera
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))

os.system('cls' if os.name == 'nt' else 'clear') # for clearing terminal screen

print('''\n
================================ BOOTING-UP =================================
\n''')

while True:
ret, frame = cap.read()
image_np = np.array(frame)

input_tensor = tf.convert_to_tensor(np.expand_dims(image_np, 0),


dtype=tf.float32)
detections = detect_fn(input_tensor)

num_detections = int(detections.pop('num_detections'))
detections = {key: value[0, :num_detections].numpy()
for key, value in detections.items()}
detections['num_detections'] = num_detections

# detection_classes should be ints.


detections['detection_classes'] =
detections['detection_classes'].astype(np.int64)
label_id_offset = 1
image_np_with_detections = image_np.copy()

vz_utils.visualize_boxes_and_labels_on_image_array(
image_np_with_detections,
detections['detection_boxes'],
detections['detection_classes']+label_id_offset,
detections['detection_scores'],
category_index,
use_normalized_coordinates=True,
max_boxes_to_draw=5,
min_score_thresh=.5,
agnostic_mode=False)

cv2.imshow('SLDP-Training-Model', cv2.resize(image_np_with_detections, (800,


600)))

if cv2.waitKey(1) & 0xFF == ord('q'):


print('''\n
================================ EXITING
================================= \n''')
cap.release()
cv2.destroyAllWindows()
break

You might also like