WIP: Add automatic content detection - #675
Conversation
|
|
||
| content_path = db.Column(db.String(200), nullable=False) | ||
| match_path = db.Column(db.String(200), nullable=False) | ||
| confidence = db.Column(db.Float, nullable=False) |
There was a problem hiding this comment.
FYI, I keep both phash and dhash seprate, it's good to have them both, right now the confidence is computed as (2*lower_of_two + higher_of_two)/3 - but we can tweak that later
| conn.execute(stmt) | ||
|
|
||
|
|
||
| class PackageContentDetection(db.Model): |
There was a problem hiding this comment.
Also here i don't see a field for dataset name, which below is called match_dataset.
| created_at = db.Column(db.DateTime, nullable=False, default=datetime.datetime.utcnow) | ||
|
|
||
| hashes = db.relationship("ContentDetectionDatasetEntryHash", back_populates="dataset_entry", | ||
| lazy="dynamic", cascade="all, delete, delete-orphan") |
There was a problem hiding this comment.
Maybe somewhat important, but reference hashes store orientation/flip too, e.g. [rot180_flip] - useful to know perhaps?
|
General thought: I was thinking about the usability on the website btw, for new releases. My thinking of the flow is like this:
|
|
Makes sense. That's the purpose of |
…r non-unique indexing to match (#676)
2 participants
Fixes #483