CODEWITHSUNDEEP
tensorflowcomputer-visionobject-detection
ma.mehralian
ma.mehralian

Reputation: 1285

Using TensorFlow Object Detection API with LSTM on a video

I am trying to track (by detection) objects on a video. The problem is that detected objects' label changed over frames of the video. I believe using RNNs (e.g., LSTMs) may help to make labels more stable but I don't have any idea how to use the frozen model of my object detector (MobilenetV2+SSD) as input for an LSTM layer and train the layer.

Upvotes: 0

Views: 987

Answers (1)

John
John

Reputation: 68

you can try this https://github.com/tensorflow/models/tree/master/research/lstm_object_detection. It implementation from Tensorflow mobile video object detection implementation proposed in the following paper: Mobile Video Object Detection with Temporally-Aware Feature Maps (CVPR 2018).

http://openaccess.thecvf.com/content_cvpr_2018/papers/Liu_Mobile_Video_Object_CVPR_2018_paper.pdf

Help you for guideline.

Upvotes: 3

Related Questions