Improve Object Detection using "Object Detection API with TensorFlow 2"
Budget: €30 – €250 EUR
I have trained a model using "Object Detection API with TensorFlow 2" ( https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf2.md ).
A result image is given in the attachment (test_001.jpg). The problem is that not enough objects are found. There are obvious objects (the green ellipses) which are not detected (no surrounding blue rectangle).
The task is to modify the model_current.config to get better results. A description what I have done and what to do is given in ReadMeForFL.txt .
The whole data including my training data and my calculated model can be dowloaded:
https://sendgb.com/lZRJa03rsSB
Once more:
- The task is to change the model_current.config to improve the results. Especially obvious objects should be detected. (The work flow should be unchanged.)
- For example, looking at test_001.jpg at least the marked objects in test_001_AdditionalObjectsToBeDetected01.jpg should be detected additionally.
The code (model_current.config) has to be delivered. Payment will only be done if the task is fully completed.
A result image is given in the attachment (test_001.jpg). The problem is that not enough objects are found. There are obvious objects (the green ellipses) which are not detected (no surrounding blue rectangle).
The task is to modify the model_current.config to get better results. A description what I have done and what to do is given in ReadMeForFL.txt .
The whole data including my training data and my calculated model can be dowloaded:
https://sendgb.com/lZRJa03rsSB
Once more:
- The task is to change the model_current.config to improve the results. Especially obvious objects should be detected. (The work flow should be unchanged.)
- For example, looking at test_001.jpg at least the marked objects in test_001_AdditionalObjectsToBeDetected01.jpg should be detected additionally.
The code (model_current.config) has to be delivered. Payment will only be done if the task is fully completed.