ImportError: cannot import name 'eval_pb2' from 'object_detection.protos' about models

[Gaurav061003]

how to solve this issue please tell me

[laxmareddyp]

Hi @Gaurav061003 ,

Could you please provide the reproducible code and share the exact repo which you are using for training the models if possible.

Thanks

[Gaurav061003]

##Reproducible code
WORKSPACE_PATH = 'Tensorflow/workspace'
SCRIPTS_PATH = 'Tensorflow/scripts'
APIMODEL_PATH = 'Tensorflow/models'
ANNOTATION_PATH = WORKSPACE_PATH+'/annotations'
IMAGE_PATH = WORKSPACE_PATH+'/images'
MODEL_PATH = WORKSPACE_PATH+'/models'
PRETRAINED_MODEL_PATH = WORKSPACE_PATH+'/pre-trained-models'
CONFIG_PATH = MODEL_PATH+'/my_ssd_mobnet/pipeline.config'
CHECKPOINT_PATH = MODEL_PATH+'/my_ssd_mobnet/'

1. Create Label Map
   labels = [{'name':'Hello', 'id':1},
   {'name':'Yes', 'id':2},
   {'name':'No', 'id':3},
   {'name':'Thanks', 'id':4},
   {'name':'I Love You', 'id':5},
   ]
   with open(ANNOTATION_PATH + '\label_map.pbtxt', 'w') as f:
   for label in labels:
   f.write('item { \n')
   f.write('\tname:'{}'\n'.format(label['name']))
   f.write('\tid:{}\n'.format(label['id']))
   f.write('}\n')

2. Create TF records
   import os
   ​
   def create_tf_record(image_dir, annotation_path, output_path):
   os.system(f"python {SCRIPTS_PATH}/generate_tfrecord.py -x {image_dir} -l {annotation_path}/label_map.pbtxt -o {output_path}")
   print(f"Successfully created the TFRecord file: {output_path}")
   ​

Example usage:

train_image_dir = os.path.join(IMAGE_PATH, 'train')
test_image_dir = os.path.join(IMAGE_PATH, 'test')
train_output_path = os.path.join(ANNOTATION_PATH, 'train.record')
test_output_path = os.path.join(ANNOTATION_PATH, 'test.record')
​
create_tf_record(train_image_dir, ANNOTATION_PATH, train_output_path)
create_tf_record(test_image_dir, ANNOTATION_PATH, test_output_path)
​
import os
import urllib.request
import tarfile

def download_pretrained_model(model_name, model_dir):
model_url = f'http://download.tensorflow.org/models/object_detection/tf2/20200711/{model_name}.tar.gz'
model_path = os.path.join(model_dir, f"{model_name}.tar.gz")

# Download the model
urllib.request.urlretrieve(model_url, model_path)

# Extract the downloaded file
with tarfile.open(model_path, 'r:gz') as tar:
    tar.extractall(model_dir)

# Remove the compressed file
os.remove(model_path)

print(f"Download and extraction of {model_name} complete.")

Example usage:

pretrained_model_name = 'ssd_mobilenet_v2_fpnlite_320x320_coco17_tpu-8'
PRETRAINED_MODEL_PATH = 'Tensorflow/workspace/pre-trained-models' # Assuming this is your predefined path
download_pretrained_model(pretrained_model_name, PRETRAINED_MODEL_PATH)

4. Copy Model Config to Training Folder
   CUSTOM_MODEL_NAME = 'my_ssd_mobnet'
   !mkdir {'Tensorflow\workspace\models\'+CUSTOM_MODEL_NAME}

5. Update Config For Transfer Learning
   import tensorflow as tf
   from google.protobuf import text_format
   from object_detection.utils import config_util
   from object_detection.protos import pipeline_pb2

##exact repo
C:\Users\RAVI\RealTimeObjectDetection-main\Tensorflow\models\research\object_detection

[laxmareddyp]

Hi @Gaurav061003,

Support for the older codebase has been discontinued. Could you kindly utilize the official Object Detection models provided by TensorFlow Model Garden.

I strongly suggest utilizing the TensorFlow Official Model Garden to circumvent issues related to outdated code commonly found in research codebases. Unlike the research repositories, the Official Model Garden is consistently updated and aligned with the latest changes in TensorFlow and other libraries.

Hope you understanding and Happy coding.

Thanks

[Gaurav061003]

but im using the updated version of [official Object Detection

[zhangj726]

hi, I met similar problem that 'cannot import module from 'object_detection.protos', have you solved it? @Gaurav061003
Thanks!