Aalen_Project_details_short paper on master thesis

[xp1632]

Expected Input and Output:

1. Hybrid Input - Text:

• This part is easy, we have a lot existing LLM model that deal with text ---> text
• text --> code, text --> use manual, text --> documentation, these only require different datasets for fine-tuning

---

2. Hybrid Input - Marked Image

• Text description could be not specific enough for certain tasks
• It's more intuitive if the end-user can send an image they want to process with certain marks

---

3. Output

• The output could be code, user manual, documentation
• or we can also provide a visual workflow so the user can further understand it

[xp1632]

Small Publication of master thesis in microscopy journals

• we want a small publication to tell the microscopy field that now they can also use ImageJ in Chaldene and Jupyter Notebook
• I'll collaborate with Dr. Miriam and have a meeting with Tim, Miriam early to set the tone for the publication

[xp1632]

Expected Input and Output:

1. Hybrid Input - Text:

• This part is easy, we have a lot existing LLM model that deal with text ---> text
• text --> code, text --> use manual, text --> documentation, these only require different datasets for fine-tuning

2. Hybrid Input - Marked Image

• Text description could be not specific enough for certain tasks
• It's more intuitive if the end-user can send an image they want to process with certain marks

3. Output

• The output could be code, user manual, documentation
• or we can also provide a visual workflow so the user can further understand it

• While here's the draft thoughts, I've summarized the technical direction in #75
• Also the topic we are looking for is Multimodal Large Language Model