Granted Patent B2

US 12,501,084 · App. 18/371,852 · Granted Dec 16, 2025

Inventors: Vignesh V. Menon (Klagenfurt am Wörthersee, AT); Hadi Amirpour (Klagenfurt am Wörthersee, AT); Christian Timmerer (Klagenfurt am Wörthersee, AT)

Assignee: Bitmovin GmbH

H04N21/2343H04N21/23418H04N21/2187

Abstract

Claims (31)

1 . A method for efficient two-pass encoding comprising:

extracting low-complexity features of a video segment;

predicting, by a neural network, an optimized constant rate factor (CRF) for the video segment using the low-complexity features;

receiving, by the neural network, a bitrate set, a resolution set, and a framerate set, wherein predicting the optimized CRF is further based on a target bitrate from the bitrate set, a resolution from the resolution set, and a framerate from the framerate set; and

encoding the video segment with the optimized CRF at a target bitrate.

2 . The method of claim 1 , wherein the extracting low-complexity features of the video segment comprises computing for each of a plurality of resolutions and framerates one or both of a spatial complexity metric and a temporal complexity metric.

3 . The method of claim 1 , wherein the low-complexity features comprise at least a Discrete Cosine Transform (DCT)-energy-based spatial feature and a DCT-energy-based temporal feature.

4 . The method of claim 1 , wherein the extracting the low-complexity features of the video segment comprises a first-pass capped variable bitrate encoding.

5 . The method of claim 1 , wherein the neural network comprises a shallow network having as few as two hidden layers.

6 . The system for efficient two-pass encoding comprising:

a processor; and

a memory comprising program instructions executable by the processor to cause the processor to implement:

a feature extraction module configured to extract a low-complexity feature from a video segment;

a neural network configured to predict an optimized constant rate factor (CRF) as a function of the low-complexity feature of the video segment and a target bitrate; and

an encoder configured to encode the video segment using the optimized CRF at the target bitrate,

wherein the neural network is trained to predict the optimized CRF as a function of the target bitrate for each of a set of resolutions and each of a set of framerates.

7 . The system of claim 6 , wherein the low-complexity feature comprises at least a Discrete Cosine Transform (DCT)-energy-based spatial feature and a DCT-energy-based temporal feature.

8 . The system of claim 6 , wherein the low-complexity feature comprises one or both of a spatial complexity metric and a temporal complexity metric.

9 . The system of claim 6 , wherein the neural network is trained to determine a minimum CRF and a maximum CRF for a target codec and a target bitrate.

10 . The system of claim 6 , wherein the neural network comprises an input layer, two or more hidden layers, and an output layer.

11 . A distributed computing system comprising:

a distributed database configured to store a plurality of video segments, a neural network, a plurality of bitrate ladders, and a codec; and

one or more processors configured to:

extract low-complexity features of a video segment;

predict, by the neural network, an optimized constant rate factor (CRF) for the video segment using the low-complexity features; and

encode the video segment with the optimized CRF at a target bitrate,

wherein the neural network is trained to predict the optimized CRF as a function of the target bitrate for each of a set of resolutions and each of a set of framerates.

12 . The system of claim 11 , wherein the low-complexity feature comprises at least a Discrete Cosine Transform (DCT)-energy-based spatial feature and a DCT-energy-based temporal feature.

13 . The system of claim 11 , wherein the low-complexity feature comprises one or both of a spatial complexity metric and a temporal complexity metric.

14 . The system of claim 11 , wherein the neural network is trained to determine a minimum CRF and a maximum CRF for a target codec and a target bitrate.

15 . The system of claim 11 , wherein the neural network comprises an input layer, two or more hidden layers, and an output layer.

Assignments (1)

Continuity (2)

References Cited (43)