Add LSTM at the beginning

[RayDu111]

Search before asking

• I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

Hi, my data includes the time information. So I tried to add LSTM in yolo to extract the time information. I want to concat the image and LSTM result, then put it to the layer 0 which is 'Conv' in yolov5s.yaml. But I have no idea which layer or API presents the image. Because only layer 0 could get the image as the input which is 'Conv' in yolov5s.yaml. Could you please give me some advice? Thank you.

Additional

No response

[pderrenger]

👋 Hello @RayDu111, thank you for your interest in YOLOv5 🚀! It sounds like you're working on an exciting project involving time-series data and YOLOv5 integration.

If this is a 🐛 Bug Report, please include a minimum reproducible example (MRE) that we can run to better understand and debug your issue.

If this is a custom training or implementation ❓ Question, please share additional details such as your proposed modifications, relevant code snippets, and error messages (if any). Providing this information will help us better assist you.

To ensure your environment is properly set up, please verify that you have Python>=3.8.0 installed along with all the required dependencies. You can do this by cloning the YOLOv5 repository, navigating to the folder, and installing the requirements using pip.

YOLOv5 is compatible with various environments, and you can run it on Jupyter Notebooks, Google Colab, cloud setups like AWS or GCP, or locally using Docker or installed dependencies. Let us know how you're running your code so we can assist further.

An Ultralytics engineer will review your request and provide personalized assistance soon! 😊

[RayDu111]

I had work it out. I added a Identity mapping layer which is layer 0. So I get the API of image. Thank you.

[pderrenger]

@RayDu111 great to hear that you resolved it by adding an identity mapping layer! If you have further questions or need assistance with any modifications, feel free to reach out. Thanks for contributing to the YOLO community!