Imageomics / Imageomics/bioclip-vector-db

Demo User Feedbacks

Open
#38 0 comments 0 reactions 1 assignee Claimed by @NetZissou View on GitHub
documentation Nice-to-have
Dominant language
Python
Stars
1
Forks
1
PR merge metrics
No merged PRs in 30d

Description

# BioCLIP Image Search Demo User Feedback

## UI Improvements

#41

**Add Navigation Instructions:** Implement tooltips or a visible "Help" legend to clarify image interaction controls:
* *Click* → Enlarge image
* *Right-click* → Go to the next image
* *Left-click* → Go to the previous image

**Rename UI Labels:**
* Update the label "Number of nearest neighbors" to: **"Number of similar images to return (nearest neighbors)"** to make it more descriptive.

* **Address "Search Depth" Complexity:** The current "Search Depth (nprobe)" control is too technical. Consider renaming this to a friendlier term (e.g., "Search Scope" or "Precision Level") or adding a simple "Low/Medium/High" abstraction layer for common users.

## Documentation

#42

**Create "How it Works" Guide:**

Draft a document explaining the mechanics of the image search.

**Visuals Required:** The documentation must include graphics/diagrams to explain technical concepts to non-experts, specifically:
* Cluster partitions (how data is grouped).
* Distance calculation (how similarity is measured).
* Ranking (how the top results are chosen).

## Functional Enhancements

#40

**Metadata Integration:**

* **Backend:** Update the FAISS server or Image Server to return metadata along with the image results.
* **Frontend:** Display this metadata alongside the search results in the app gallery.

**BioCLIP-2 Prediction:**
* Add a feature to run BioCLIP-2 prediction on the *input* image.
* **Performance:** This should ideally run as a concurrent process so it does not block the search results.

## Architecture & Refactoring

#39

**Backend Extraction (Model Service):**
* Refactor the application to move the embedding process off the frontend.
* Create a dedicated **Model Service Server** that provides endpoints for:
1. Generating Embeddings.
2. Running Predictions.

Goal of this is to gradually reduce frontend calculation load by offloading tasks to this service. Frankly after the embedding offload, the Gradio app is pretty much just the front-end.

Contributor guide

No contributing guide indexed for this repository

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.