The MERL Shopping Dataset contains 106 videos, each of which is a sequence ~2 minutes long. Each video contains several instances of the following 6 actions:
- Reach To Shelf - (reach to shelf)
- Retract From Shelf - (retract hand from shelf)
- Hand In Shelf - (extended period with hand in the shelf)
- Inspect Product - (inspect product while holding it in hand)
- Inspect Shelf - (look at shelf while not touching and not reaching for the shelf)
- None of the above
- wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
- chmod u+x Miniconda3-latest-Linux-x86_64.sh
- ./Miniconda3-latest-Linux-x86_64.sh
- cd miniconda3/
- conda list
- source .bashrc
- conda create -n tf python=3.7 anaconda
- conda activate tf
- conda install -c anaconda tensorflow-gpu
- conda install -c conda-forge opencv
- Download and extract MERL Shopping Dataset - ftp://ftp.merl.com/pub/tmarks/MERL_Shopping_Dataset/
- Download Videos_MERL_Shopping_Dataset.zip and unzip it.
- Create train, val and test folders inside the unzipped folder for Videos_MERL_Shopping_Dataset.
- Move 1st 60 videos (1_1_crop to 20_3_crop) to the train folder.
- Move the next 18 videos (21_1_crop to 26_3_crop) to the val folder
- Move the remaining 28 videos (27_1_crop to 41_2_crop) to the test folder
- Then run the following command on the terminal "./convert_video_to_images.sh Videos_MERL_Shopping_Dataset"
- Repeat the above steps 2-6 similarly for Labels_MERL_Shopping_Dataset and then run Generate_output_pickle_files.ipynb.
- Activate the conda environment and run ActionRecognition.py on terminal or ActionRecognition.ipynb to run in the jupyter environment
- In order to run ResNet Model run Resnet.py
- Singh, Bharat, et al. "A multi-stream bi-directional recurrent neural network for fine-grained action detection." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2016.