Back to results
US 2020/0218888 A1Published application
Target Re-Identification
AssigneeVision Semantics Limited
InventorsShaogang GONG, Xiatian ZHU, Hanxiao WANG, Xu LANG
Life of this patent
Estimating the term…
Loading patent…
Loading patent…
AssigneeVision Semantics Limited
InventorsShaogang GONG, Xiatian ZHU, Hanxiao WANG, Xu LANG
Estimating the term…
A computer implemented method and system for training a machine to identify a target within video data, the method comprising the steps of providing a training data set including identified labelled targets within video data having the same target within different video views. Generating, using a learning model, a bounding box action policy for determining required adjustments to a bounding box around a target in the video data by: generating a bounding box around a labelled target within a first view of the video data. Converting the target bounded by the bounding box to a quantitative representation. Determining a matching level between the quantitative representation and a quantitative representation of a further labelled target within the video data from a second view different to the first view. Looping the following steps one or more times, the looped steps comprising: using the bounding box action policy to determine an action to change the bounding box. Applying the determined action to change the bounding box to a new bounding box. Converting the target bounded by the new bounding box to a new quantitative representation. Determining a further matching level between the new quantitative representation and the quantitative representation of the labelled target within the video data from the second view. Applying a reward to the learning model to adjust the bounding box action policy based on an improvement in the matching level.
A computer implemented method of training a machine to identify a target within video data, the method comprising the steps of:
The computer implemented method of claim, wherein the feature vectors are 1-D high dimensional feature maps, 2-D feature maps, or 3-D volumes.
The computer implemented method according to any previous claim, wherein the first view and the second view contain a target with the same label.
The computer implemented method according to any previous claim, wherein the bounding box is rectangular.
The computer implemented method according to any previous claim, wherein the action to change the bounding box is any one or more of:
The computer implemented method according to any previous claim, wherein the reward is a positive reward if the matching level improves and wherein the reward is a negative reward if the matching level decreases.
The computer implemented method according to any previous claim, wherein the reward is applied using a reward function.
The computer implemented method according to any previous claim, wherein the method is implemented by a neural network.
The computer implemented method according to any previous claim, wherein the loop ends when a criteria is reached.
The computer implemented method according to any previous claim further comprising the step of:
The computer implemented method according to any previous claim, wherein the loop is ended after a predetermined number of loops.
The computer implemented method according to any previous claim, wherein the first view and the second view come from different sources of data and/or video cameras.
A system comprising:
Dependencies are inferred from the claim text.