Class VideoCLIP<T>
- Namespace
- AiDotNet.Video.Understanding
- Assembly
- AiDotNet.dll
VideoCLIP model for video-text understanding and retrieval.
public class VideoCLIP<T> : NeuralNetworkBase<T>, INeuralNetworkModel<T>, INeuralNetwork<T>, IFullModel<T, Tensor<T>, Tensor<T>>, IModel<Tensor<T>, Tensor<T>, ModelMetadata<T>>, IModelSerializer, ICheckpointableModel, IParameterizable<T, Tensor<T>, Tensor<T>>, IFeatureAware, IFeatureImportance<T>, ICloneable<IFullModel<T, Tensor<T>, Tensor<T>>>, IGradientComputable<T, Tensor<T>, Tensor<T>>, IJitCompilable<T>, IInterpretableModel<T>, IInputGradientComputable<T>, IDisposableType Parameters
TThe numeric type used for calculations (e.g., float, double).
- Inheritance
-
VideoCLIP<T>
- Implements
- Inherited Members
- Extension Methods
Remarks
For Beginners: VideoCLIP learns to understand both videos and text descriptions in a shared "embedding space" where similar concepts are close together.
Key capabilities:
- Video-to-Text Search: Find text descriptions that match a video
- Text-to-Video Search: Find videos that match a text query
- Zero-Shot Classification: Classify videos into categories without training
- Video Captioning: Generate descriptions for videos
- Video Question Answering: Answer questions about video content
The model creates embeddings (numerical representations) for both videos and text that can be compared using similarity measures. Videos and their corresponding descriptions will have similar embeddings.
Technical Details: - Contrastive learning on video-text pairs - Temporal transformer for video understanding - Text transformer for language understanding - Joint embedding space with cosine similarity - Pre-trained on large-scale video-text datasets
Reference: Xu et al., "VideoCLIP: Contrastive Pre-training for Zero-shot Video-Text Understanding" EMNLP 2021.
Constructors
VideoCLIP(NeuralNetworkArchitecture<T>, int, int, int, int, double, string?, string?)
Initializes a new instance of the VideoCLIP class.
public VideoCLIP(NeuralNetworkArchitecture<T> architecture, int numFrames = 32, int embeddingDim = 512, int textMaxLength = 77, int vocabSize = 49408, double temperature = 0.07, string? vocabPath = null, string? mergesPath = null)Parameters
architectureNeuralNetworkArchitecture<T>The neural network architecture configuration.
numFramesintNumber of video frames to process.
embeddingDimintDimension of the shared embedding space.
textMaxLengthintMaximum text sequence length.
vocabSizeintVocabulary size for text encoding.
temperaturedoubleTemperature for softmax scaling.
vocabPathstringOptional path to CLIP vocabulary JSON file for production tokenization.
mergesPathstringOptional path to CLIP BPE merges file for production tokenization.
Remarks
For Production Use: Provide vocabPath and mergesPath to use proper CLIP tokenization. Download these files from HuggingFace's openai/clip-vit-base-patch32 repository: - vocab.json: Token vocabulary mapping - merges.txt: BPE merge rules
For Testing: Omit vocabPath and mergesPath to use a simple test tokenizer.
Properties
SupportsTraining
Gets whether training is supported.
public override bool SupportsTraining { get; }Property Value
Methods
ComputeSimilarity(Tensor<T>, Tensor<T>)
Computes similarity between video and text embeddings.
public double ComputeSimilarity(Tensor<T> videoEmbedding, Tensor<T> textEmbedding)Parameters
videoEmbeddingTensor<T>Video embedding.
textEmbeddingTensor<T>Text embedding.
Returns
- double
Similarity score (higher = more similar).
ComputeSimilarityMatrix(List<Tensor<T>>, List<Tensor<T>>)
Computes video-text similarity matrix for a batch.
public Tensor<T> ComputeSimilarityMatrix(List<Tensor<T>> videoFramesBatch, List<Tensor<T>> textsBatch)Parameters
videoFramesBatchList<Tensor<T>>Batch of videos [B, T, C, H, W].
textsBatchList<Tensor<T>>Batch of text token IDs [B, SeqLen].
Returns
- Tensor<T>
Similarity matrix [B, B] where (i,j) is similarity between video i and text j.
CreateNewInstance()
Creates a new instance of the same type as this neural network.
protected override IFullModel<T, Tensor<T>, Tensor<T>> CreateNewInstance()Returns
- IFullModel<T, Tensor<T>, Tensor<T>>
A new instance of the same neural network type.
Remarks
For Beginners: This creates a blank version of the same type of neural network.
It's used internally by methods like DeepCopy and Clone to create the right type of network before copying the data into it.
DeserializeNetworkSpecificData(BinaryReader)
Deserializes network-specific data that was not covered by the general deserialization process.
protected override void DeserializeNetworkSpecificData(BinaryReader reader)Parameters
readerBinaryReaderThe BinaryReader to read the data from.
Remarks
This method is called at the end of the general deserialization process to allow derived classes to read any additional data specific to their implementation.
For Beginners: Continuing the suitcase analogy, this is like unpacking that special compartment. After the main deserialization method has unpacked the common items (layers, parameters), this method allows each specific type of neural network to unpack its own unique items that were stored during serialization.
EncodeText(Tensor<T>)
Encodes text into an embedding vector.
public Tensor<T> EncodeText(Tensor<T> tokenIds)Parameters
tokenIdsTensor<T>Token IDs [SeqLen] or [B, SeqLen].
Returns
- Tensor<T>
Text embedding [EmbeddingDim] or [B, EmbeddingDim].
Remarks
For Beginners: This converts text (as token IDs) into a numerical vector. Text with similar meaning will have similar embeddings.
EncodeVideo(Tensor<T>)
Encodes a video into an embedding vector.
public Tensor<T> EncodeVideo(Tensor<T> videoFrames)Parameters
videoFramesTensor<T>Input video [T, C, H, W] or [B, T, C, H, W].
Returns
- Tensor<T>
Video embedding [EmbeddingDim] or [B, EmbeddingDim].
Remarks
For Beginners: This converts a video into a numerical vector (embedding). Videos with similar content will have similar embeddings.
GetModelMetadata()
Gets the metadata for this neural network model.
public override ModelMetadata<T> GetModelMetadata()Returns
- ModelMetadata<T>
A ModelMetaData object containing information about the model.
InitializeLayers()
Initializes the layers of the neural network based on the architecture.
protected override void InitializeLayers()Remarks
For Beginners: This method sets up all the layers in your neural network according to the architecture you've defined. It's like assembling the parts of your network before you can use it.
Predict(Tensor<T>)
Makes a prediction using the neural network.
public override Tensor<T> Predict(Tensor<T> input)Parameters
inputTensor<T>The input data to process.
Returns
- Tensor<T>
The network's prediction.
Remarks
For Beginners: This is the main method you'll use to get results from your trained neural network. You provide some input data (like an image or text), and the network processes it through all its layers to produce an output (like a classification or prediction).
SerializeNetworkSpecificData(BinaryWriter)
Serializes network-specific data that is not covered by the general serialization process.
protected override void SerializeNetworkSpecificData(BinaryWriter writer)Parameters
writerBinaryWriterThe BinaryWriter to write the data to.
Remarks
This method is called at the end of the general serialization process to allow derived classes to write any additional data specific to their implementation.
For Beginners: Think of this as packing a special compartment in your suitcase. While the main serialization method packs the common items (layers, parameters), this method allows each specific type of neural network to pack its own unique items that other networks might not have.
TextToVideoRetrieval(string, List<Tensor<T>>, int)
Retrieves the most similar videos to a text query.
public List<(int VideoIndex, double Similarity)> TextToVideoRetrieval(string query, List<Tensor<T>> videoEmbeddings, int topK = 10)Parameters
querystringText query describing the desired video.
videoEmbeddingsList<Tensor<T>>Pre-computed video embeddings.
topKintNumber of results to return.
Returns
- List<(int ClassIndex, double Probability)>
List of (videoIndex, similarity) pairs, sorted by similarity.
Train(Tensor<T>, Tensor<T>)
Trains the neural network on a single input-output pair.
public override void Train(Tensor<T> input, Tensor<T> expectedOutput)Parameters
inputTensor<T>The input data.
expectedOutputTensor<T>The expected output for the given input.
Remarks
This method performs one training step on the neural network using the provided input and expected output. It updates the network's parameters to reduce the error between the network's prediction and the expected output.
For Beginners: This is how your neural network learns. You provide: - An input (what the network should process) - The expected output (what the correct answer should be)
The network then:
- Makes a prediction based on the input
- Compares its prediction to the expected output
- Calculates how wrong it was (the loss)
- Adjusts its internal values to do better next time
After training, you can get the loss value using the GetLastLoss() method to see how well the network is learning.
UpdateParameters(Vector<T>)
Updates the network's parameters with new values.
public override void UpdateParameters(Vector<T> parameters)Parameters
parametersVector<T>The new parameter values to set.
Remarks
For Beginners: During training, a neural network's internal values (parameters) get adjusted to improve its performance. This method allows you to update all those values at once by providing a complete set of new parameters.
This is typically used by optimization algorithms that calculate better parameter values based on training data.
VideoToTextRetrieval(Tensor<T>, List<string>, int)
Retrieves the most similar text descriptions for a video.
public List<(string Text, double Similarity)> VideoToTextRetrieval(Tensor<T> videoFrames, List<string> candidateTexts, int topK = 10)Parameters
videoFramesTensor<T>Input video frames.
candidateTextsList<string>List of candidate text descriptions.
topKintNumber of results to return.
Returns
- List<(string Label, double Confidence)>
List of (text, similarity) pairs, sorted by similarity.
ZeroShotClassify(Tensor<T>, List<string>)
Performs zero-shot video classification.
public List<(string ClassName, double Probability)> ZeroShotClassify(Tensor<T> videoFrames, List<string> classTexts)Parameters
videoFramesTensor<T>Input video frames.
classTextsList<string>List of class descriptions (e.g., "a video of cooking").
Returns
- List<(string Label, double Confidence)>
Probability distribution over classes.
Remarks
For Beginners: This classifies videos without any training on those specific categories. Simply provide text descriptions of each class (like "a video of someone running"), and the model will determine which description best matches the video.