search

EdjeElectronics / TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10

How to train a TensorFlow Object Detection Classifier for multiple object detection on Windows
Apache License 2.0
2.92k stars 1.3k forks source link

Improving a pre-trained tensorflow object detection model #477

Open ArefAz opened 4 years ago

ArefAz commented 4 years ago

I want to use tensorflow for detecting cars in an embedded system, so I tried ssd_mobilenet_v2 and it actually did pretty well for me, except for some specific car types which are not very common and I think that is why the model does not recognize them. I have a dataset of these cases and I want to improve the model by fine-tuning it. I should also note that I need a .tflite file because I'm using tflite_runtime in python. I followed these instructions https://github.com/EdjeElectronics/TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10 and I could train the model and reached a reasonable loss value. I then used export_tflite_ssd_graph.py in the object detection API to build inference_graph from the trained model. Afterwards I used toco tool to build a .tflite file out of it.

But here is the problem, after I've done all that; not only the model did not improve, but now it does not detect any cars. I got confused and do not know what is the problem, I searched a lot and did not find any tutorial about doing what I need to do. They just added a new object to a model and then exported it, which I tried and I was successful doing that. I also tried to build a .tflite file without training the model and directly from the Tensorflow detection model zoo and it worked fine. So I think the problem has something to do with the training process. Maybe I am missing something there.

I would like to know whether it is possible to improve detection for a specific class. In my case I want to detect cars in my application but the current models do not detect some specific cars that are not very common in the streets. What is the solution? Should I download the coco dataset and add my data to it then train it again? Or is there any other way to do it?

Another thing that I did not find in documents is that whether is it possible to "add" a class to the current classes of an object detection model. For example, let's assume the mobilenet ssd v2 detects 90 different object classes, I would like to add another class so that the model detects 91 different classes instead of 90 classes. As far as I understand and tested after doing transfer learning using object detection API, I could only detect the objects that I had in my dataset and the old classes will be gone. So how do I do what I explained?

mouhannadali commented 3 years ago

any update here? I would like also to improving a pre-trained TensorFlow object detection model