A easy and fast custom Object Detection installer based on Tensorflow Object Detection API for Linux (Needs Root)
Clone the Repo: git clone https://github.com/BySuxax/Easy_AI_Detection
Step by Step:
-
type in console: InstallAllLibarys.bash and awnser the questions
-
when the gedit window open: type in the number of clases you want to detect at num_classes: and save it!
-
when LabelImage window opened; Open the train folder (yourDir/models/research/object_detection/train) with "Open Dir" button:
-
When Window has opened with the images start drawing rectangels(with w), type class names and save it. repeat it for every image in train!
-
Then open the test folder and select the test folder(yourDir/models/research/object_detection/test) with "Open Dir" button:
-
When Window has opened with the images start drawing rectangels(with w), type class names and save it. repeat it for every image in test!
-
when gedit window opened, Add to: AND SAVE!
def class_text_to_int(row_label):
if row_label == 'YOUR_LABEL1':
return 1
#Past Here
else:
None
for every label a:
elif row_label == 'YOUR_LABEL_NAME':
return 2
-
please enter in the following shema. Per item you want to detect: AND SAVE!
id rising up starts at 1 name = name you labeled it
item{ id: 1 name: 'YOR_FIRST_LABEL' }
-
Now the training should start. Wait till the loss avrage is under 0.5/0.4 then press Strg+C
-
run in Terminal first cd ../../../ and then . Create_Graph.bash LAST_SAVED_CHECKPOINT_NUM
-
Now you can use the Test1.py file to test the training. If you want to us it in a outher project you can use the file but you have to change the sys.path.appand("YOUR path to the Easy_AI_Detection.py file in the repo")
HAVE FUN!
######################################################################### QUELLE: https://github.com/datitran/raccoon_dataset, https://github.com/tensorflow/tensorflow, https://github.com/tzutalin/labelImg, https://www.youtube.com/watch?v=COlbP62-B-U&list=PLQVvvaa0QuDcNK5GeCQnxYnSSaar2tpku&index=1
LICENSES: 1. xml_to_css.py,generate_tfrecord.py ./LICENSE_raccoon_dataset 2. tensorflow ./models/LICENSE 3. labelImg ./models/research/object_detection/labelImg/LICENSE 4. Easy_AI_Detection ./LICENSE
Thanks to: datitran, tensorflow, tzutalin, sentdex