Table 2 Algorithm for query processing.

From: AI-driven video summarization for optimizing content retrieval and management through deep learning techniques

Input

Input query

Output

Extract Query Features:\(Q_{features} = ResNet50\left( {query_{image} } \right)\)

Extract Frame Features:\(F_{features} = ResNet50\left( {frame} \right)\)

Compute Similarity Scores: \(S_{f} = \frac{{Q_{features } . F_{features} }}{{\left| {\left| {Q_{features } } \right|} \right| \left| {\left| { F_{features} } \right|} \right|}}\)

Sort Frames by Relevance:

\(R = sorted\left( {R.Key = lambdax:x\left[ 1 \right],reverse = True} \right)\)

Select Top N Frames: \(N = {\text{min}}\left( {N.len\left( R \right)} \right)\)

Return Top Keyframe:\(K = \left[ {R\left[ i \right]\left[ 0 \right]for i in range \left( N \right)} \right]\)

Back to article page