Skip to content

Feature 6: Fish Species Detection Before Freshness Analysis #161

Description

@saidai-bhuvanesh

Description

Different fish species (e.g. Salmon, Tuna, Rohu, Catla, Mrigal) have distinct decay patterns, gill structures, and optical properties. A one-size-fits-all freshness model limits accuracy. This issue proposes implementing a Fish Species Detection pipeline prior to running the freshness models.

When an image is uploaded, it will first pass through a species classifier to identify the fish type. Once detected, the app will calibrate and load species-specific parameters and weights for the freshness models. In demo/offline modes, a visual color/texture and aspect ratio classifier will simulate this detection. The UI will present a clear status stage ("Detecting species...") and display a species card with common name, scientific name, and habitat tags.

Technical Implementation Details

  1. Species Classifier: Define species classification logic in backend/main.py.
  2. Parity in Frontend: Port species mapping using color profiles and shapes to fusionInference.js.
  3. Dynamic Calibrations: Map thresholds and storage hours dynamically based on species.
  4. UI Status: Enhance ScannerPage camera states to include a loader for species detection.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions