Importing a Tiny YoloV3 Model from Keras

May 3, 2022

A LITTLE HISTORY

In 2016ย Redmon, Divvala, Girschick and Farhadiย revolutionized object detection with a paper titled:ย You Only Look Once: Unified, Real-Time Object Detection.ย In the paper they introduced a new approach to object detection โ€” The feature extraction and object localization were unified into a single monolithic block. Furthermore โ€” the localization and classification heads were also united. Their single-stage architecture, namedย YOLOย (You Only Look Once) results in a very fast inference time.

DATASET USED TO TRAIN THE MODEL :ย COCO (MICROSOFT COMMON OBJECTS IN CONTEXT)

The MS COCO (Microsoft Common Objects in Context) dataset is a large-scale object detection, segmentation, key-point detection, and captioning dataset. The dataset consists of 328K images.

The dataset has annotations for panoptic: full scene segmentation, with 80 thing categories (such as person, bicycle, elephant) and a subset of 91 stuff categories (grass, sky, road).

THE ARCHITECTURE

Inspired by ResNet and FPN (Feature-Pyramid Network) architectures, Tiny YOLO-V3 feature extractor, called Darknet-19 and 2 prediction heads (like FPN) โ€” each processing the image at a different spatial compression.

IMPORTING FROM HDF5 SAVE KERAS FORMAT A TINY YOLOV3

HAIBALย Library can import all HDF5 saved file from Keras Library.ย  As application, we import a Tiny Yolov3 exemple fromย GitHubย and use it on ourย HAIBALย LabVIEW DeepLearning Library.

In the below images, we predict the same result (boxes) as the Keras Model.

On the left the image processed by Keras and on the right the same image processed with our HAIBAL library (city)

On the left the image processed by Keras and on the right the same image processed with our HAIBAL library (Open Space)

On the left the image processed by Keras and on the right the same image processed with our HAIBAL library (Street)

Theย Tiny YoloV3 pretrained modelย exemple will be proposedย  in the HAIBAL library to permit our community to use and modify it.

Software needed to run Tiny YoloV3 model

  • LabVIEW 2020 (or latest)
  • HAIBAL Deep Learning development module
  • Recommended Vision development module 2020 (or latest) to capture pictures and display the overlays
TIGR1.2.5 release notes

TIGR1.2.5 release notes

All release notes are available at this page .Download link Release NotesV1.2.5 Date of releaseย 30 November 2023 Enhancements: Palette Removal: Removed the LabVIEW version palette. New Palettes...

read more
TIGR1.2 release notes

TIGR1.2 release notes

All release notes are available at this page .Download link Release NotesV1.2 Date of releaseย 24 Octobre 2023 Features Bugs/Improve Comment Update palette We changed the operation of the palette...

read more
TIGR1.1.1 release notes

TIGR1.1.1 release notes

All release notes are available at this page .Download link Release NotesV1.1.1 Date of releaseย 30 Septembre 2023 Features Bugs/Improve Comment Update palette โ€“ ROI selector work now in any...

read more