Team 16's code submission is a zipped copy of https://github.com/pratyushghosh/CS4243_Mini_Project. In order to test the final model attempted by Team 16, open the following Colaboratory notebook and connect to a GPU-acceleration-enabled runtime (ideally with high RAM): https://colab.research.google.com/drive/1rGDgCCjdV3McRFNhnAIeUGX4Tc-4f_94?usp=sharing. - Run all the cells under the sections: - Clone Repository and Install Dependencies - Build OpenPose (this step could take several minutes, depending upon the available RAM) - Upload all test images into the directory /content/cs4243_miniproject_team16/test_images - Run all the cells under the sections: - Run Model - Generate Confusion Matrix Note that in order for the confusion matrix to be generated correctly, the correct label for each test image should be contained as a substring within its filename, eg. A0170549Y_20220904_threat.jpg Team 16's baseline model and training code can be found at: https://colab.research.google.com/github/pratyushghosh/CS4243_Mini_Project/blob/main/Baseline_Image_Classification.ipynb. It can also be seen inside the code submission, in the folder training_code_and_baseline. Team 16's pose-analysis model and training code can be found at: https://colab.research.google.com/drive/1069awpfZSgsOh6v0tSdHPgrmZPnC61zA?usp=sharing This is referred to as Model B in the final two-step model. It can also be seen inside the code submission, in the folder training_code_and_baseline. Team 16's gun-detection model is a trained model based on an open-source project. Its training code can be found at https://github.com/RunzeXU/AI-detection-weapons.