keineahnung2345

In https://github.com/rushter/MLAlgorithms/blob/master/mla/pca.py#L48, the variance ratio of a component is calculated as the square of its eigenvalue divided by the sum of all squared eigenvalues. But according to https://stats.stackexchange.com/questions/31908/what-is-percentage-of-variance-in-pca, the proportion...

Ref: Canny(image, threshold1, threshold2[, edges[, apertureSize[, L2gradient]]]) -> edges

Handle zero pad & Allow bg to train RPN

24

Handle zero pad: - fpn_classifier_graph: first use target_class_ids to find out the samples which are zero padding and finally zero out the result - mrcnn_class_loss_graph: use target_class_ids to find out...

[octree] use PCL_ERROR instead of assert

2

In release build, `assert` is ignored, this PR replaces them with `PCL_ERROR`.

[feat] gicp: add convergence criteria and correspondence estimation

3

Attempt to fix https://github.com/PointCloudLibrary/pcl/issues/3329.

[registration] GICP alternates optimizing trans and rot part

2

Ref: https://github.com/PointCloudLibrary/pcl/issues/5180

The link given in README.md is dead now. This PR updates the link of "CTPN+CRNN implementation" to a new one.

layers.py: Keras 2.2 use layer._inbound_nodes, layer._outbound_nodes instead of layer.inbound_nodes, layer.outbound_nodes. utils.py conv_utils.normalize_data_format should be replaced with K.normalize_data_format in keras 2.2. Ref: ekholabs/keras-contrib@0dac2da#diff-0970ecd5f755557421e7ba7716bdfd6a https://github.com/apple/coremltools/blob/master/coremltools/converters/keras/_topology2.py

[filters] FrustumCulling: asymmetrical Field of View

4

Attemp to implement the feature mentioned in https://github.com/PointCloudLibrary/pcl/issues/5422