LogoLogo
HomeDocsAPI & SDKsProjectsForumStudio
  • Welcome
    • Featured Machine Learning Projects
      • Getting Started with the Edge Impulse Nvidia TAO Pipeline - Renesas EK-RA8D1
      • Smart City Traffic Analysis - NVIDIA TAO + Jetson Orin Nano
      • ROS 2 Pick and Place System - Arduino Braccio++ Robotic Arm and Luxonis OAK-D
      • Optimize a cloud-based Visual Anomaly Detection Model for Edge Deployments
      • Rooftop Ice Detection with Things Network Visualization - Nvidia Omniverse Replicator
      • Surgery Inventory Object Detection - Synthetic Data - Nvidia Omniverse Replicator
      • NVIDIA Omniverse - Synthetic Data Generation For Edge Impulse Projects
      • Community Guide – Using Edge Impulse with Nvidia DeepStream
      • Computer Vision Object Counting - Avnet RZBoard V2L
      • Gesture Appliances Control with Pose Detection - BrainChip AKD1000
      • Counting for Inspection and Quality Control - Nvidia Jetson Nano (TensorRT)
      • High-resolution, High-speed Object Counting - Nvidia Jetson Nano (TensorRT)
    • Prototype and Concept Projects
      • Renesas CK-RA6M5 Cloud Kit - Getting Started with Machine Learning
      • TI CC1352P Launchpad - Getting Started with Machine Learning
      • OpenMV Cam RT1062 - Getting Started with Machine Learning
      • Getting Started with Edge Impulse Experiments
  • Computer Vision Projects
    • Workplace Organizer - Nvidia Jetson Nano
    • Recyclable Materials Sorter - Nvidia Jetson Nano
    • Analog Meter Reading - Arduino Nicla Vision
    • Creating Synthetic Data with Nvidia Omniverse Replicator
    • SonicSight AR - Sound Classification with Feedback on an Augmented Reality Display
    • Traffic Monitoring - Brainchip Akida
    • Multi-camera Video Stream Inference - Brainchip Akida
    • Industrial Inspection Line - Brainchip Akida
    • X-Ray Classification and Analysis - Brainchip Akida
    • Inventory Stock Tracker - FOMO - BrainChip Akida
    • Container Counting - Arduino Nicla Vision
    • Smart Smoke Alarm - Arduino Nano 33
    • Shield Bot Autonomous Security Robot
    • Cyclist Blind Spot Detection - Himax WE-I Plus
    • IV Drip Fluid-Level Monitoring - Arduino Portenta H7
    • Worker PPE Safety Monitoring - Nvidia Jetson Nano
    • Delivered Package Detection - ESP-EYE
    • Bean Leaf Disease Classification - Sony Spresense
    • Oil Tank Measurement Using Computer Vision - Sony Spresense
    • Object Counting for Smart Industries - Raspberry Pi
    • Smart Cashier with FOMO - Raspberry Pi
    • PCB Defect Detection with Computer Vision - Raspberry Pi
    • Bicycle Counting - Sony Spresense
    • Counting Eggs with Computer Vision - OpenMV Cam H7
    • Elevator Passenger Counting - Arduino Nicla Vision
    • ESD Protection using Computer Vision - Seeed ReComputer
    • Solar Panel Defect Detection - Arduino Portenta H7
    • Label Defect Detection - Raspberry Pi
    • Dials and Knob Monitoring with Computer Vision - Raspberry Pi
    • Digital Character Recognition on Electric Meter System - OpenMV Cam H7
    • Corrosion Detection with Computer Vision - Seeed reTerminal
    • Inventory Management with Computer Vision - Raspberry Pi
    • Monitoring Retail Checkout Lines with Computer Vision - Renesas RZ/V2L
    • Counting Retail Inventory with Computer Vision - Renesas RZ/V2L
    • Pose Detection - Renesas RZ/V2L
    • Product Quality Inspection - Renesas RZ/V2L
    • Smart Grocery Cart Using Computer Vision - OpenMV Cam H7
    • Driver Drowsiness Detection With FOMO - Arduino Nicla Vision
    • Gastroscopic Image Processing - OpenMV Cam H7
    • Pharmaceutical Pill Quality Control and Defect Detection
    • Deter Shoplifting with Computer Vision - Texas Instruments TDA4VM
    • Smart Factory Prototype - Texas Instruments TDA4VM
    • Correct Posture Detection and Enforcement - Texas Instruments TDA4VM
    • Visual Anomaly Detection with FOMO-AD - Texas Instruments TDA4VM
    • Surface Crack Detection and Localization - Texas Instruments TDA4VM
    • Surface Crack Detection - Seeed reTerminal
    • Retail Image Classification - Nvidia Jetson Nano
    • SiLabs xG24 Plus Arducam - Sorting Objects with Computer Vision and Robotics - Part 1
    • SiLabs xG24 Plus Arducam - Sorting Objects with Computer Vision and Robotics - Part 2
    • Object Detection and Visualization - Seeed Grove Vision AI Module
    • Bike Rearview Radar - Raspberry Pi
    • Build a Self-Driving RC Vehicle - Arduino Portenta H7 and Computer Vision
    • "Bring Your Own Model" Image Classifier for Wound Identification
    • Acute Lymphoblastic Leukemia Classifier - Nvidia Jetson Nano
    • Hardhat Detection in Industrial Settings - Alif Ensemble E7
    • Motorcycle Helmet Identification and Traffic Light Control - Texas Instruments AM62A
    • Import a Pretrained Model with "Bring Your Own Model" - Texas Instruments AM62A
    • Product Inspection with Visual Anomaly Detection - FOMO-AD - Sony Spresense
    • Visual Anomaly Detection in Fabric using FOMO-AD - Raspberry Pi 5
    • Car Detection and Tracking System for Toll Plazas - Raspberry Pi AI Kit
    • Visual Anomaly Detection - Seeed Grove Vision AI Module V2
    • Object Counting with FOMO - OpenMV Cam RT1062
    • Visitor Heatmap with FOMO Object Detection - Jetson Orin Nano
    • Vehicle Security Camera - Arduino Portenta H7
  • Audio Projects
    • Occupancy Sensing - SiLabs xG24
    • Smart Appliance Control Using Voice Commands - Nordic Thingy:53
    • Glass Window Break Detection - Nordic Thingy:53
    • Illegal Logging Detection - Nordic Thingy:53
    • Illegal Logging Detection - Syntiant TinyML
    • Wearable Cough Sensor and Monitoring - Arduino Nano 33 BLE Sense
    • Collect Data for Keyword Spotting - Raspberry Pi Pico
    • Voice-Activated LED Strip - Raspberry Pi Pico
    • Snoring Detection on a Smart Phone
    • Gunshot Audio Classification - Arduino Nano 33 + Portenta H7
    • AI-Powered Patient Assistance - Arduino Nano 33 BLE Sense
    • Acoustic Pipe Leakage Detection - Arduino Portenta H7
    • Location Identification using Sound - Syntiant TinyML
    • Environmental Noise Classification - Nordic Thingy:53
    • Running Faucet Detection - Seeed XIAO Sense + Blues Cellular
    • Vandalism Detection via Audio Classification - Arduino Nano 33 BLE Sense
    • Predictive Maintenance Using Audio Classification - Arduino Nano 33 BLE Sense
    • Porting an Audio Project from the SiLabs Thunderboard Sense 2 to xG24
    • Environmental Audio Monitoring Wearable - Syntiant TinyML - Part 1
    • Environmental Audio Monitoring Wearable - Syntiant TinyML - Part 2
    • Keyword Spotting - Nordic Thingy:53
    • Detecting Worker Accidents with Audio Classification - Syntiant TinyML
    • Snoring Detection with Syntiant NDP120 Neural Decision Processor - Arduino Nicla Voice
    • Recognize Voice Commands with the Particle Photon 2
    • Voice Controlled Power Plug with Syntiant NDP120 (Nicla Voice)
    • Determining Compressor State with Audio Classification - Avnet RaSynBoard
    • Developing a Voice-Activated Product with Edge Impulse's Synthetic Data Pipeline
    • Enhancing Worker Safety using Synthetic Audio to Create a Dog Bark Classifier
  • Predictive Maintenance and Defect Detection Projects
    • Predictive Maintenance - Nordic Thingy:91
    • Brushless DC Motor Anomaly Detection
    • Industrial Compressor Predictive Maintenance - Nordic Thingy:53
    • Anticipate Power Outages with Machine Learning - Arduino Nano 33 BLE Sense
    • Faulty Lithium-Ion Cell Identification in Battery Packs - Seeed Wio Terminal
    • Weight Scale Predictive Maintenance - Arduino Nano 33 BLE Sense
    • Fluid Leak Detection With a Flowmeter and AI - Seeed Wio Terminal
    • Pipeline Clog Detection with a Flowmeter and AI - Seeed Wio Terminal
    • Refrigerator Predictive Maintenance - Arduino Nano 33 BLE Sense
    • Motor Pump Predictive Maintenance - Infineon PSoC 6 WiFi-BT Pioneer Kit + CN0549
    • BrickML Demo Project - 3D Printer Anomaly Detection
    • Condition Monitoring - Syntiant TinyML Board
    • Predictive Maintenance - Commercial Printer - Sony Spresense + CommonSense
    • Vibration Classification with BrainChip's Akida
    • AI-driven Audio and Thermal HVAC Monitoring - SeeedStudio XIAO ESP32
  • Accelerometer and Activity Projects
    • Arduino x K-Way - Outdoor Activity Tracker
    • Arduino x K-Way - Gesture Recognition for Hiking
    • Arduino x K-Way - TinyML Fall Detection
    • Posture Detection for Worker Safety - SiLabs Thunderboard Sense 2
    • Hand Gesture Recognition - OpenMV Cam H7
    • Arduin-Row, a TinyML Rowing Machine Coach - Arduino Nicla Sense ME
    • Fall Detection using a Transformer Model – Arduino Giga R1 WiFi
    • Bluetooth Fall Detection - Arduino Nano 33 BLE Sense
    • Monitor Packages During Transit with AI - Arduino Nano 33 BLE Sense
    • Smart Baby Swing - Arduino Portenta H7
    • Warehouse Shipment Monitoring - SiLabs Thunderboard Sense 2
    • Gesture Recognition - Bangle.js Smartwatch
    • Gesture Recognition for Patient Communication - SiLabs Thunderboard Sense 2
    • Hospital Bed Occupancy Detection - Arduino Nano 33 BLE Sense
    • Porting a Posture Detection Project from the SiLabs Thunderboard Sense 2 to xG24
    • Porting a Gesture Recognition Project from the SiLabs Thunderboard Sense 2 to xG24
    • Continuous Gait Monitor (Anomaly Detection) - Nordic Thingy:53
    • Classifying Exercise Activities on a BangleJS Smartwatch
  • Air Quality and Environmental Projects
    • Arduino x K-Way - Environmental Asthma Risk Assessment
    • Gas Detection in the Oil and Gas Industry - Nordic Thingy:91
    • Smart HVAC System with a Sony Spresense
    • Smart HVAC System with an Arduino Nicla Vision
    • Indoor CO2 Level Estimation - Arduino Portenta H7
    • Harmful Gases Detection - Arduino Nano 33 BLE Sense
    • Fire Detection Using Sensor Fusion and TinyML - Arduino Nano 33 BLE Sense
    • AI-Assisted Monitoring of Dairy Manufacturing Conditions - Seeed XIAO ESP32C3
    • AI-Assisted Air Quality Monitoring - DFRobot Firebeetle ESP32
    • Air Quality Monitoring with Sipeed Longan Nano - RISC-V Gigadevice
    • Methane Monitoring in Mines - Silabs xG24 Dev Kit
    • Smart Building Ventilation with Environmental Sensor Fusion
    • Sensor Data Fusion with Spresense and CommonSense
    • Water Pollution Detection - Arduino Nano ESP32 + Ultrasonic Scan
    • Fire Detection Using Sensor Fusion - Arduino Nano 33 BLE Sense
  • Novel Sensor Projects
    • 8x8 ToF Gesture Classification - Arduino RP2040 Connect
    • Food Irradiation Dose Detection - DFRobot Beetle ESP32C3
    • Applying EEG Data to Machine Learning, Part 1
    • Applying EEG Data to Machine Learning, Part 2
    • Applying EEG Data to Machine Learning, Part 3
    • Liquid Classification with TinyML - Seeed Wio Terminal + TDS Sensor
    • AI-Assisted Pipeline Diagnostics and Inspection with mmWave Radar
    • Soil Quality Detection Using AI and LoRaWAN - Seeed Sensecap A1101
    • Smart Diaper Prototype - Arduino Nicla Sense ME
    • DIY Smart Glove with Flex Sensors
    • EdgeML Energy Monitoring - Particle Photon 2
    • Wearable for Monitoring Worker Stress using HR/HRV DSP Block - Arduino Portenta
  • Software Integration Demos
    • Azure Machine Learning with Kubernetes Compute and Edge Impulse
    • ROS2 + Edge Impulse, Part 1: Pub/Sub Node in Python
    • ROS2 + Edge Impulse, Part 2: MicroROS
    • Using Hugging Face Datasets in Edge Impulse
    • Using Hugging Face Image Classification Datasets with Edge Impulse
    • Edge Impulse API Usage Sample Application - Jetson Nano Trainer
    • MLOps with Edge Impulse and Azure IoT Edge
    • A Federated Approach to Train and Deploy Machine Learning Models
    • DIY Model Weight Update for Continuous AI Deployments
    • Automate the CI/CD Pipeline of your Models with Edge Impulse and GitHub Actions
    • Deploying Edge Impulse Models on ZEDEDA Cloud Devices
Powered by GitBook
On this page
  • Introduction
  • Story
  • Why Synthetic Data?
  • Data-Centric Workflow
  • Requirements
  • Hardware and Driver Setup
  • Experiment Setup and Data Generation
  • Data Distribution and Model Building
  • Edge Impulse: Data Annotation and Model Building
  • Testing of Object Detection Models with Real Objects
  • Conclusion
  • Reference

Was this helpful?

Edit on GitHub
Export as PDF
  1. Computer Vision Projects

Creating Synthetic Data with Nvidia Omniverse Replicator

Learn how to generate photorealistic images in Nvidia Omniverse Replicator and build an object detection model using Edge Impulse.

PreviousAnalog Meter Reading - Arduino Nicla VisionNextSonicSight AR - Sound Classification with Feedback on an Augmented Reality Display

Last updated 1 year ago

Was this helpful?

Created By:

Public Project Link: |

Introduction

Building an accurate and high-quality object detection model can be tricky since it requires a large dataset. Sometimes, data can be hard to acquire, or not diverse enough to train a robust model. Synthetic data offers an alternative to generating well-represented datasets to build a quality model. By applying domain randomization, we developed photorealistic datasets, trained a neural network, and validated the model using real-world data. To create a diverse dataset, we created a variety of simulated environments with randomized properties: changing lighting conditions, camera position, and material textures. We also show that synthetic, randomized datasets can help generalize a model to adapt to the real-world environment.

Story

Why Synthetic Data?

Computer vision tasks such as classification, object detection, and segmentation require a large-scale dataset. Data collected from some real-world applications tend to be narrow and less diverse, often collected from a single environment, and sometimes data is unchanged and stays the same. In addition, data collected from a single field tends to have fewer examples of tail-end scenarios and rare events, and we cannot easily replicate these situations in the real world.

Consequently, models trained in a single domain are brittle and often fail when deployed in another environment; thus, it requires another training cycle to adapt to the new environment. It raises the question, how can we efficiently and cheaply collect generalized data across several domains? A simple effective solution is Domain Randomization, which varies the texture and colour of the foreground object, the background image, the number of lights in the scene, the pose of the lights, and the camera position etc. Domain randomization can further improve the variability in the texture of synthetic data of rare events generated in the simulator.

The purpose of domain randomization is to provide enough simulated variability at training time such that at test time the model is able to generalize to real-world data.” - Tobin et al, Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World, 2017

Nvidia Replicator enables us to perform Domain Randomization. The Replicator is one module within the Omniverse family, and it offers tools and workflow to generate data for various computer vision and non-visual tasks. The Replicator is a highly interoperable tool that integrates with over 40+ modelling/rendering applications across different verticals. The seamless integration is possible thanks to Pixar's Universal Scene Description (USD), which serves as a protocol for various applications such as Blender, 3DMax, Maya, Revit, C4D etc., to work with the Nvidia Replicator.

Data-Centric Workflow

Traditional machine learning workflow is often model-centric, focusing more on the model's development by iteratively improving the algorithm design, etc. In this project, we chose the Data-centric approach, where we fixed the model and iteratively improved the quality of the generated dataset. This approach is more robust since we know our model is as good as the dataset. This method hence systematically changes the dataset performance on an AI task. At its core, it is thinking about ML in terms of data, not the model.

Requirements

  • Edge Impulse Studio

  • USB Webcam (Logitech C920 is used in this project, but any camera attached and detected by the Edge Impulse CLI will work)

Hardware and Driver Setup

Nvidia Omniverse Replicator is a computation-intensive application requiring a moderate-size GPU and decent RAM. My hardware setup consists of 32GB RAM, 1TB storage space and an 8GB GPU with an Intel i9 processor.

The application can run on both Windows and Linux operating systems. For this experiment, we used Ubuntu 20.04 LTS distro, given Ubuntu 18.04 is no longer supported by Nvidia Omniverse as of November 2022. In addition, we selected the appropriate Nvidia driver, v510.108.03 and installed it on a Linux machine.

Experiment Setup and Data Generation

The environment for the experiment consists of movable and immovable objects (dynamic and static positioning objects). The immovable object consists of Lights, a Table and two Cameras. The movable objects are the cutlery, which is a spoon, fork and knife. We will use domain randomization to alter the properties of some of the movable and immovable objects. Assets which include objects and scenes are represented in the Replicator as USD.

Every object in Omniverse Replicator is represented as USD. A 3D model file with varying extensions such as .obj, .fbx, and .glif can be imported into the Replicator using Nvidia Omniverse's CAD Importer extension. The extension converts the 3D files into USD. We imported our assets (table, knife, spoon, and fork) into the simulator by specifying the path of the assets.

Lightning plays a crucial role in data generation. There are different built-in lighting types in the Nvidia replicator. We choose two rectangular lights and a dome light since they give us better lighting options and capabilities for generating photorealistic images. The rectangular light emulates light generated from a panel, and the dome light lets you dynamically lighten the entire scene. We randomized some light parameters such as temperature and intensity, and both parameters were sampled from a normal distribution. In addition, the scale parameter was sampled from a uniform distribution while keeping the rotation and position of the lights fixed.

python
# Lightning setup for Rectangular light and Dome light 

def rect_lights(num=2):
    lights = rep.create.light(
    light_type="rect",
    temperature=rep.distribution.normal(6500, 500),
    intensity=rep.distribution.normal(0, 5000),
    position=(-131,150,-134),
    rotation=(-90,0,0),
    scale=rep.distribution.uniform(50, 100),
    count=num
    )
    return lights.node

def dome_lights(num=1):
    lights = rep.create.light(
    light_type="dome",
    temperature=rep.distribution.normal(6500, 500),
    intensity=rep.distribution.normal(0, 1000),
    position=(0,0,0),
    rotation=(270,0,0),
    count=num
    )
return lights.node

We fixed the position and rotation, selected the tabletop materials, chose an additional Mahogany material, and alternated the material in the data generation process.

python 
# Import and position the table object

def table():
    table = rep.create.from_usd(TABLE_USD, semantics=[('class', 'table')])

    with table:
        rep.modify.pose(
            position=(-135.39745, 0, -140.25696),
            rotation=(0,-90,-90),
        )
    return table

To improve our dataset's quality further, we chose two cameras of different resolutions, which we strategically positioned in various locations within the scene. In addition, we varied the position of the cameras in a different version of the data generation process.

# Multiple setup cameras and attach it to render products
camera = rep.create.camera(focus_distance=focus_distance, focal_length=focal_length, position=cam_position, rotation=cam_rotation, f_stop=f_stop)
camera2 = rep.create.camera(focus_distance=focus_distance2, focal_length=focal_length2, position=cam_position2, rotation=cam_rotation, f_stop=f_stop)

# Will render 1024x1024 images and 512x512 images
render_product  = rep.create.render_product(camera1, (1024, 1024))
render_product2  = rep.create.render_product(camera2, (512, 512))

Finally, for the movable objects, which include a knife, spoon, and fork, we ensure that these objects can only translate within the bound of the table. So we chose a bounding position where the objects were expected to translate and rotate with the table. We sampled position and rotation from a uniform distribution while maintaining the number of movable objects generated at each iteration to be five.

python 
# Define randomizer function for CULTERY assets.
    def cutlery_props(size=5):
        instances = rep.randomizer.instantiate(rep.utils.get_usd_files(current_cultery), size=size, mode='point_instance')

        with instances:
            rep.modify.pose(
                position=rep.distribution.uniform((-212, 76.2, -187), (-62, 76.2, -94)),
                rotation=rep.distribution.uniform((-90,-180, 0), (-90, 180, 0)),
            )
        return instances.node

At this juncture, we have instantiated all objects in our scene. We can now run the randomizer to generate 50 images at each synthetic generation cycle.

python
# Register randomization
with rep.trigger.on_frame(num_frames=50):
    rep.randomizer.table()
    rep.randomizer.rect_lights(1)
    rep.randomizer.dome_lights(1)
    rep.randomizer.cutlery_props(5)

# Run the simulation graph
rep.orchestrator.run()

To ensure we generated photorealistic images, we switched to RTXinteractive(Path Tracing) mode, which gave high-fidelity renderings.

Data Distribution and Model Building

Following the data-centric philosophy, We generated three versions of the dataset. The first version, V1, consists of generated images normal to the camera position, and V2 represents images generated at an angle of 60 degrees to the camera position with a mahogany table top. V3 comprises images normal to the camera position while the cutlery were suspended in space.

This resulted in the following Generated Dataset:

Edge Impulse: Data Annotation and Model Building

Data Labeler
Data Annotation

We carefully annotated each dataset version and trained using the Yolov5 object detection model. We tried a couple of input sizes ranging from 320, 512 and 1024 pixels before settling on 320 pixels. Edge Impulse provided an excellent version control system for models, which enabled us to track model performance across different dataset versions and hyperparameters.

Create Impulse
Generate Feature

Testing of Object Detection Models with Real Objects

We used the Edge Impulse CLI tool to evaluate the model's accuracy by downloading, building and running the model locally. A Logitech C920 webcam streamed the live video of objects on a table from 50 cm to 80 cm from the camera. The position of the camera remains fixed during the experiment. The clips below show that the trained model does not generalize well to real-world objects. Thus we needed to improve the model by uploading, annotating and training the model with the V2 dataset.

We observed improved model performance when trained with the V2 dataset. The model could identify various objects distinctly, although the model failed when we changed the objects' orientations. Thus, we trained the model with the remaining V3 dataset to mitigate these issues and increase other hyperparameters, such as epochs from 500 to 2000. We also tested the performance of our object detector on real objects with different background textures, and the model performed well in these conditions.

After several cycles of iterating over various hyperparameters, we got a model that generalizes well across different orientations.

The core idea behind the data-centric approach to solving ML problems is to create more data around the failure points of the model. We improved the model by iteratively improving the data generation, especially in areas where the model had previously failed.

Conclusion

In this work, we learned how the Domain Randomization approach helps generate quality and well-generalized datasets for an object detection task. We also demonstrated the effectiveness of data-centric machine learning workflows in improving model performance. Although this work is restricted to visual problems, we can extend domain randomization to other sensors such as lidar, accelerometer, and ultrasonic sensors as well.

Reference

We wanted to replicate the work done by Louis Moreau, but this time using synthetic data rather than real data. The project aims to demonstrate how to build and deploy an Edge Impulse object detection model using synthetic datasets generated by Nvidia Omniverse Replicator. The Replicator is an Nvidia Omniverse extension that provides a means of generating physically accurate synthetic data.

We manually uploaded the generated images to Edge Impulse Studio, where we annotated the dataset into different classes, but there is also an Omniverse plugin that can be installed to automate the upload process. Information about the plugin installation process and usage .

object detection
Nvidia Omniverse Replicator
can be found here
Project on Edge Impulse
Introduction to Replicator
Introduction to USD
Telsa AI Day
Domain Randomization for Transferring Deep Neural Networks
Understanding Domain Randomization for SIM-TO-REAL Transfer
George Igwegbe
GitHub
Edge Impulse
Andrej Karpathy's presentation - (source: Tesla AI Day, 2021)
Domain Randomization for Transferring Deep Neural Networks - source: Tobin et al, 2017)
Data generation and model building workflow
Data generation process
Data Distribution of different items
V1 - Normal to the object
V2 - Angled to the object
V3 - Normal to the object and object suspended in space
Generated Dataset - V3
Generated Dataset - V3
Version Control in Edge Impulse
V1 failure - model failed to identify objects
V2 success - model can identify objects
V2 failure - model failed to identify objects in different orientations
V3 success - model can identify objects in different orientations
V3 success - model can identify different materials