FLAVROptions
Options & Configuration · AiDotNet.Video.Options
Configuration options for FLAVR, flow-agnostic video
representations for fast frame interpolation (Kalluri et al., WACV 2023).
Inheritance: ModelOptions → NeuralNetworkOptions → FLAVROptions
For Beginners
FLAVR never estimates motion. It stacks four frames into a short clip, runs 3D (space and time) convolutions over it in a U-Net, and directly paints the in-between frame or frames.
How It Works
The defaults are the release’s (tarun005/FLAVR: model/FLAVR_arch.py, model/resnet_3D.py, config.py, main.py): the
unet_18 encoder (R3D-18 with feature gating, no batch norm), four input frames and one output frame for 2x
interpolation; Deep returns the unet_34 encoder. Setting NumOutputFrames to 3 or 7 gives
the release’s 4x and 8x models.
Constructors
| Constructor | Summary |
|---|---|
FLAVROptions() |
Initializes a new instance with default values. |
FLAVROptions(FLAVROptions other) |
Initializes a new instance by copying from another instance. |
Properties
| Property | Summary |
|---|---|
OnnxModelOptions OnnxOptions { get; set; } |
Gets or sets the ONNX runtime options. |
double Beta1 { get; set; } |
Gets or sets Adam’s first-moment decay (0.9). |
double Beta2 { get; set; } |
Gets or sets Adam’s second-moment decay (0.99). |
double LearningRate { get; set; } |
Gets or sets Adam’s initial learning rate (2e-4). |
double PlateauFactor { get; set; } |
Gets or sets the factor the rate is multiplied by when the validation loss plateaus (0.5). |
int BaseChannels { get; set; } |
Gets or sets the encoder’s first width; the four stages are 1, 2, 4 and 8 times it (64). |
int NumInputFrames { get; set; } |
Gets or sets the context frames, half before and half after the gap (4). |
int NumOutputFrames { get; set; } |
Gets or sets the evenly spaced frames synthesized in the gap (1 for 2x; 3 for 4x; 7 for 8x). |
int PlateauPatience { get; set; } |
Gets or sets the epochs without improvement before the rate drops (5). |
int[] LayerBlocks { get; set; } |
Gets or sets the residual blocks in each of the encoder’s four stages (unet_18: 2, 2, 2, 2). |
string ModelPath { get; set; } |
Gets or sets the path to the ONNX model file. |
Methods
| Method | Summary |
|---|---|
FLAVROptions Deep() |
The release’s unet_34 encoder (R3D-34 block counts). |