: Extracting the portrait image from a document while it is being moved or held in different lighting.
: Features 1,000 unique mock documents with artificially generated faces to preserve privacy. MIDV-250-CS.mp4
: Testing how well text fields can be read when the camera isn't perfectly still or the background is distracting. 3. Dataset Variations : Extracting the portrait image from a document
: The unique identifier for a specific document type. For example, in the MIDV-500 dataset, there are 50 document types, each associated with 10 videos (500 total). : Represents the capturing condition "Clutter"
: Represents the capturing condition "Clutter" . In this condition, the scene and background are intentionally filled with unrelated objects to challenge document localization algorithms. mp4 : The standard video file format used for these clips. 2. Context and Usage
The "MIDV" family has evolved to include more challenging data: : The original dataset with 500 clips.
: Extracting the portrait image from a document while it is being moved or held in different lighting.
: Features 1,000 unique mock documents with artificially generated faces to preserve privacy.
: Testing how well text fields can be read when the camera isn't perfectly still or the background is distracting. 3. Dataset Variations
: The unique identifier for a specific document type. For example, in the MIDV-500 dataset, there are 50 document types, each associated with 10 videos (500 total).
: Represents the capturing condition "Clutter" . In this condition, the scene and background are intentionally filled with unrelated objects to challenge document localization algorithms. mp4 : The standard video file format used for these clips. 2. Context and Usage
The "MIDV" family has evolved to include more challenging data: : The original dataset with 500 clips.