2.2. Training for Object Detection

This guide trains a neural network model to object detection of Human Face in the Open Images Dataset V4 on GPU server.

2.2.1. Preparation

The Open Images V4 dataset is available from official website, but original dataset is too large for the tutorial. So, We provide reduced dataset.

Blueoil supports 2 formats for object detection.

  • OpenImagev4 format
  • DeLTA-Mark format

Note: Please see the detail in Prepare training dataset

You can download subset of Open Images V4 from our server.

$ wget https://s3-ap-northeast-1.amazonaws.com/leapmind-public-storage/datasets/openimages_face.tgz
$ tar xf openimages_face.tgz

This dataset consists of 2866 Human Face images and 5170 annotation boxes.

2.2.2. Generate a configuration file

Generate your model configuration file interactively by running blueoil.sh init command.

$ ./blueoil.sh init

This is an example of initialization.

#### Generate config ####
your model name ():  objectdetection
choose task type  object_detection
choose network  LMFYoloQuantize
choose dataset format  OpenImagesV4
training dataset path:  {dataset_dir}
set validataion dataset? (if answer no, the dataset will be separated for training and validation by 9:1 ratio.)  no
batch size (integer):  16
image size (integer x integer):  224x224
how many epochs do you run training (integer):  1000
initial learning rate:  0.001
choose learning rate schedule ({epochs} is the number of training epochs you entered before):  '3-step-decay' -> learning rate decrease by 1/10 on {epochs}/3 and {epochs}*2/3 and {epochs}-1
enable data augmentation?  Yes
Please choose augmentors:  done (5 selections)
-> select Brightness, Color, FlipLeftRight, Hue, SSDRandomCrop
apply quantization at the first layer?  no

2.2.3. Train a network model

Train your model by running blueoil.sh train command with model configuration.

$ ./blueoil.sh train config/{Model name}.yml

When training is started, training log and checkpoints are generated under ./saved/{Mode name}_{TIMESTAMP} directory.

Training is running on TensorFlow backend. So you can use TensorBoard to visualize your training process.

$ ./blueoil.sh tensorboard saved/{Model name}_{TIMESTAMP} {Port}
  • Metrics / Accuracy
  • Loss, Weight Decay
  • Images / Final Detect Boxes

2.2.4. Convert training result to FPGA ready format.

Convert trained model to executable binary files for x86, ARM, and FPGA. Currently, conversion for FPGA only supports Intel Cyclone® V SoC FPGA.

$ ./blueoil.sh convert config/{Model name}.yml saved/{Mode name}_{TIMESTAMP}

blueoil.sh convert automatically executes some conversion processes.

  • Convert Tensorflow checkpoint to protocol buffer graph.
  • Optimize graph
  • Generate source code for executable binary
  • Compile for x86, ARM and FPGA

If conversion is successful, output files are generated under ./saved/{Mode name}_{TIMESTAMP}/export/save.ckpt-{Checkpoint No.}/{Image size}/output.

 ├── fpga (include preloader and FPGA configuration file)
 │   ├── preloader-mkpimage.bin
 │   └── soc_system.rbf
 ├── models
 │   ├── lib (include trained model library)
 │   │   ├── lib_arm.so
 │   │   ├── lib_fpga.so
 │   │   └── lib_x86.so
 │   └── meta.yaml (model configuration)
 ├── python
 │   ├── lmnet (include pre-process/post-process)
 │   ├── README.md
 │   ├── requirements.txt
 │   ├── run.py (inference script)
 │   └── usb_camera_demo.py (demo script for object detection)
 └── README.md

2.2.5. Run inference script on x86 Linux (Ubuntu 16.04)

  • Prepare images for inference (not included in the training dataset)

    You can find test imgaes on Creative Commons. Sample

      $ wget https://farm4.staticflickr.com/1172/1144309435_eff42ee683_o.jpg
  • Run inference script

    Explore into the output/python directory, and run run.py and inference result is saved in ./output/output.json.

    Note: If you run the script for the first time, you have to setup a python environment (2.7 or 3.5+) and required python packages.

    $ cd {output/python directory}
    $ sudo pip install -r requirements.txt  # for the first time only
    $ python run.py \
          -i {inference image path} \
          -l ../models/lib/lib_x86.so \
          -c ../models/meta.yaml

    Tips: The default thredhold for object detection is 0.05. If you find too many boxses when running demo, you can edit meta.yml and set threshold to 0.4 or 0.5 as below code.

        threshold: 0.4
  • Check inference result

        "classes": [
                "id": 0,
                "name": "Humanface"
        "date": "2019-01-25T13:15:00.928639",
        "results": [
                "file_path": "{inference_image_path}",
                "prediction": [
                        "box": [
                        "class": {
                            "id": 0,
                            "name": "Humanface"
                        "score": "0.8219463229179382"
        "task": "IMAGE.OBJECT_DETECTION",
        "version": 0.2