logiSTK is a software tool for training and testing of HOG/SVM-based classifiers. It is designed to support users in complex operations, such as the learning process for an automated classifier, in a simple and effective way. With the logiSTK software tool, even non-experts for computer vision can undertake the task of classifiers training. The tool natively supports the
logiHOG Object Detector IP core, which is integrated in the logiADAK ADAS development kit, as an actual target.
Key Features
Media Annotation: The logiSTK projects use videos (.dv, .avi, .mov, .wmv, .mpeg2, .mp4) or collections of images (.jpg, .jpeg, .bmp, .png, .tiff, .ppm, .pgm) for classification, training and testing. To annotate the media, logiSTK supports the concept of layers as collection of boxes referred to a specific object of interest (e.g. pedestrian, motorcycle, car, traffic sign, animal, cell, etc.). The tool can automatically generate boxes between the selected frames to speed up the manual annotation.
Training: logiSTK exposes a guided procedure through all training phases. The procedure relies on a training set of annotated media for positive and negative images and/or binary datasets (features previously computed and exported from the logiSTK after the training process). The guided procedure embraces fine tuning for training parameters and can display a qualitative representation of the computed classifier able to detect objects of interest in project video/images.
ROI (Region of Interest): The logiSTK can define a region on the video frames/images where the objects of interest are supposed to be. Different scales of a ROI are usually used for each video frame/image to let the classifier detecting objects at different scales (image pyramid). The logiSTK supports ROIs that apply to different media and can be used with different classifiers within the same template size. ROIs can be created manually (specify min/max scale and density), or automatically starting from camera parameters.
Performance analysis: The logiSTK enables evaluation of the classification results of a classifier. With at least one ground truth data layer and one layer with detection results (and optionally a layer containing regions to be ignored), the tool can compute standard metrics about classifier's performances, such as precision, recall, F1-score, TP, FP, and FN.
Trial version
A 15-days trial
version for Windows is available free of charge. We recommend to first-time users to download the tutorial archive as well. The tutorial and
included test projects will run you through the whole process of
training a classifier for No Entry Signs with logiSTK.
For more information about the logiSTK Software Training Kit, please contact
info@logicBRICKS.com.
|
|
|
The logiSTK software tool is sourced from our Technology Partner eVS embedded Vision Systems Srl. |