Method and system for automated behavior classification of test subjects
The method and system includes a plurality of test subject containers, each having a test subject therein. A plurality of video cameras is focused on the test subject containers to capture video of the behavior of each of the test subjects. The system and method also includes a storage system for storing the plurality of video from the cameras. The storage system may be local or cloud-based as is known in the art. The system further has one or more computers with a neural network configured to (a) retrieve the video from the storage system of the test subjects, (b) analyze the video to identify a plurality of observable behaviors in the test subjects, (c) annotate the video with the observed behavior classifications, and (d) store the annotated video in the storage system.
1. A method of automated behavior classification of a plurality of test subjects, comprising:
providing video of a test subject;
annotating the video of the test subject with a plurality of behavior classifications corresponding to a behavior of a test subject in the video at a point in time, thereby creating annotated video;
storing the annotated video;
training a neural network with the annotated video to identify a relationship of the test subject in the video with the annotated behavior classification for that test subject at that point in time in the annotated video; and
with the trained neural network, classifying behavior of a plurality of test subjects from a plurality of videos.
2. The method of claim 1 , wherein the relationship comprises a shape of the test subject.
3. The method of claim 1 , wherein the relationship comprises a motion of the test subject.
4. The method of claim 1 , wherein the relationship comprises a position of the test subject.
5. The method of claim 1 , further comprising pre-training the neural network by providing predetermined weight and bias values to the neural network.
6. The method of claim 1 , wherein the neural network comprises a convolution neural network.
7. The method of claim 6 , wherein the neural network comprises a plurality of convolution layers interspersed with a plurality of pooling layers.
8. The method of claim 6 , wherein the neural network further comprises a plurality of fully connected convolution layers.
9. The method of claim 6 , wherein the neural network further comprises a recurrent neural network layer.
10. The method of claim 9 , wherein the recurrent neural network layer comprises an LSTM layer.
11. The method of claim 10 , wherein the LSTM layer is intersperse within the fully connected convolution layers.
12. A method of automated behavior classification of a plurality of test subjects, comprising:
providing a neural network trained to identify a plurality of observable behaviors in test subjects;
providing video of a plurality of test subjects to the neural network;
with the neural network, identifying the observed behaviors of the test subjects in the video,
annotating the time the observed behavior occurred, thereby creating an annotated video; and
storing the annotated video.
13. The method of claim 12 , wherein the neural network is configured to identify a shape of the test subject.
14. The method of claim 12 , wherein the neural network is configured to identify a motion of the test subject.
15. The method of claim 12 , wherein the neural network is configured to identify a position of the test subject.
16. The method of claim 12 , wherein the annotations comprise a label indicating the observed behavior.
17. The method of claim 12 , wherein the annotations comprise a timestamp indicating the time the observed behavior occurred in the video.
18. The method of claim 12 , further comprising creating a log of a time the observed behavior occurred and a type of observed behaviors that the test subject made in the video.