Machine Learning (ML) is becoming increasingly popular and is being adopted in real-life AI applications across different industries. Creating a real-world AI application often requires extensive work with unstructured data. In fact, ML engineers and data scientists spend more than 80% of their time on data preparation and labeling and only a fraction of their time on the so-called fun stuff: reading research papers, training models, trying new architectures, tuning hyperparameters, deploying, monitoring, etc.
Since ML engineers are spending such a huge portion of their time on structuring, labeling, versioning, and debugging datasets to become AI-ready training data (aka SuperData), data labeling toolsets have become essential for building scalable AI applications. However, data labeling tools or simple data labeling editors are far away from covering all the growing needs of anyone's complex ML pipeline.
We identified 6 essential components that make data labeling tools a compelling solution for building modern AI pipelines. Namely, annotation software, AI data management and curation, integrations and security, project and quality management, and automation.
Therefore, this article is neither about simple data labeling tools using open-source software such as label studio, cvat, labelme, nor is it about specific functionalities within labeling editors such as bounding boxes, polygons, text labeling, etc.
We will cover various types of data labeling companies, along with their history and functionalities, detailed feature sets, additional AI pipeline-oriented components, and much more. We will rely on one of the most reliable software ranking marketplaces, G2, where data labeling is a separate software category within various AI software. As we indicated above, for each company/solution, we will cover the following:
- Annotation software (image annotation, video annotation, text annotation, etc.)
- AI Data Management and curation (active learning, query system, smart sampling, subset selection, data versioning, etc.)
- Integrations and security (Storage integrations, model inference API or model training integrations, etc.)
- MLOps and Automation (SDK, webhooks, orchestrations, AI-enabled labeling, model management, etc.)
- Project and quality management (Team/role management, annotation project management, and performance tracking)
- Integrated annotation services
We will be updating this article quarterly and tracking all the position changes, feature releases, large news, and announcements.
1. SuperAnnotate
Ranked as the best data labeling platform in G2, SuperAnnotate builds an end-to-end data solution with an integrated service marketplace, where they help their customers to find the right annotation team within the preferred geographic location and proficiency. The labeling teams are directly integrated into SuperAnnotate's platform and managed by their professional project managers which are reported to be one of their key strengths. The company also provides data on its platform with various labeling tools without annotation services.
Company history: Started as a Ph.D. research, SuperAnnotate was founded in 2018, primarily as an image annotation tool for semantic segmentation, but quickly gained momentum and extended into other areas of ML pipeline development. SuperAnnotate raised about $22M from investors such as P9 and Base10.
Key features: The software itself focuses on 5 key components of the AI lifecycle:
- Data labeling tools and efficient project management
- AI dataset management, data curation, and version control
- AI model management, model comparison, and versioning
- Automation and orchestration via various triggering systems (available for enterprises)
- Annotation Service Marketplace
Data labeling tools: On the labeling side, the company started with image annotation capabilities. Over the last 2 years, SuperAnnotate launched video and text annotation editors. Earlier in 2022, SuperAnnotate's platform evolved into LiDAR and audio labeling. Fasting forward, just introduced more targeted labeling data formats, such as native PDF, DICOM, etc.
MLOps: While covering different data formats, it is also important to cover other MLOps capabilities, which makes such annotation tooling companies more compelling within the entire AI lifecycle. On that end, SuperAnnotate’s capabilities become very attractive for any startup or enterprise users. More particularly, easy project management, data curation, data versioning, model management, automation, and complete SDK allow customers to automate incredibly complex AI pipelines (as reviews confirm).
Security: SuperAnnotate also offers multiple levels of data security, allowing users to store their datasets on their premises or in SuperAnnotate’s encrypted S3 buckets. The company owns several certifications, including SOC2 Type II on the software side and ISO 27001, GDPR, and HIPPA compliant.
Annotation Services: On the services side, SuperAnnotate vetted over 400 annotation service teams and allows its users to find teams in different geographies, languages, and medical experts, as well as relatively cost-efficient annotation services for easier tasks such as image classification.
G2 review summary: SuperAnnotate now carries the title of “Leader Fall 2022” and “1st Easiest To Use” on G2’s data labeling software list, which is a testament to its resolute commitment and consistency.
Out of the 75 reviews (4.9/5), most users believe that SuperAnnotate’s platform is super-user-friendly and contains all the features required for annotating unstructured data. Many also praise the platform's data management and ability to track the progress of annotation tasks across different projects while leveraging useful metrics. Users also assure they do not experience major problems with the tool at all. When faced with minor issues, the support service is always there to assist in a timely manner. There is a mention of how the upload and export of pipelines are a bit burdensome when just starting, yet the user also adds that SuperAnnotate is always adding useful functionality to reduce these issues.
Pricing: The software is FREE for up to 10 users and 50.000 images. For higher-level commitments, make sure to consult with the sales department.
2. V7
Company history: Founded in 2018, V7Labs was initially built as an image annotation tool and then extended toward model building and automation functionalities. Prior to V7Labs, the founders built a company called AIPoly, which enables the visually impaired to see and name various objects through the phone camera. The company is based in the UK and has raised around $43M.
Data labeling tools: On the business side, V7Labs focuses on visual data, helping customers to solve primarily computer vision problems. V7 also has model management and document processing systems, but based on several client feedbacks it has several bugs and fails using the automations, especially when the load is high.
Services: V7’s annotation services offer dozens of agents to label and refine images, yet it also gives users the opportunity to bring in their own labeling team to create training data or support the human-in-the-loop processes.
G2 review summary: As we write this article, the company has a 4.8/5 rating with a total of 41 reviews, earning itself the title of “2nd Easiest To Use” data labeling software on G2. However, it’s worth mentioning that almost half of the user reviews are coming from Ghana, which creates some suspicions around the validity of the reviews. Users also recall that V7's occasional tendency to lag when working with large datasets increases the number of times users spend on a specific project. A few users comment on security glitches (data becomes available to the public) and inconsistent billing computations.
Pricing: V7 offers four payment options: the Education package (which is free of charge), the Zero to AI version (for up to 10 people startups), and the Business and Pro versions.
3. Dataloop
Company history: Founded in 2017, Dataloop is an end-to-end platform covering every step from development to production with a technology that also comprises a data management and labeling platform. It has raised an estimated $50M at this time of writing.
Data labeling tools and project management: When it comes to data labeling tools, Dataloop provides a toolset for image, video, and text annotation formats. It also provides a data infrastructure for labeling various data types. It is an end-to-end platform that covers annotation (image, video, and lidar), data QA and verification, data, workforce and project management, and automation.
Security: The platform is an enterprise-ready solution committed to ensuring top-quality data organization and collaboration that's in line with key security and privacy standards across the industry.
G2 user summary: Reserving the 3rd spot on the list, Dataloop has a rating of 4.3/5 and 44 reviews. Most of the reviews are proof that Dataloop is fairly simple to use and provides good services. Many express that Dataloop’s software is very helpful in the administrative field and in getting the labeling done in a short period of time. There are some complaints about Dataloop’s constant price increase since they rise with each update, as per reviewers. Users also state that the tool’s performance occasionally slows down when working with large datasets.
Pricing: Offers a free trial, no other pricing information disclosed.
4. Keymakr
Company history: With a mission to build and shape better technology, Keymakr started as a 10-people-company back in 2015. Now, it is considered one of the top data labeling companies that offers annotation services for image, video, and document annotation, data creation, and collection.
Data labeling tools and services: Keymakr offers a wide range of services, including image, video, and document annotation, automation, dataset validation, open-source data collection, and data creation in Keymakr's dedicated studio based on specific company needs.
Security: With Keymaker your data is kept private and secure as they apply encryption, data expiration, VPN solutions, and more. Unfortunately, we did not find any documentation or youtube videos to better understand the capabilities of the platform.
G2 user summary: Keymakr is #4 on G2’s list of the best data labeling tools, with a 4.8/5 rating and 23 reviews.
As for the reviews, the users did not concentrate on the platform and primarily talked about their services. Almost all of the customer reviews are signs of Keymakr’s responsiveness, work ethic, alignment, and customer service, as many state that they respect deadlines and do not overpromise. Some users did mention that, at times, communication can be delayed because of the difference in time zones and that Keymakr’s prices are a bit higher compared to other tools.
Pricing: Keymakr offers a free trial, yet it has 3 pricing editions; Startup, Business, and Business Pro.
5. Labelbox
Company history: Labelbox is a training data platform founded in 2017. The founders were building internal tools in different companies in the aerospace industry. They saw the pain of creating image annotation tools and came together to create a company to address their pain points. Nowadays, Labelbox builds real-world AI and machine learning with software built for industrial data science teams. The startup has received $190 million in funding from Gradient Ventures, First Round Capital, Kleiner Perkins, and Andreesen Horowitz (a16z).
Data labeling tools: Similar to Dataloop, SuperAnnotate, and Labelbox also provide data labeling tools for various data types. The all-in-one platform is a foundation for users to easily build and improve training data for their AI.
Project management and services: It is designed around the following data pillars: AI-assisted labeling, data curation, data ops automation with Python SDK, workspace navigation and management, model training and diagnostics, as well as on-demand labeling services.
G2 user summary: With a 4.7/5 rating and 20 reviews, Labelbox is G2’s Leader of Fall 2022 and #5 on the list.
The user reviews indicate that Labelbox is effective and simple. The instructions are easy to follow, and many state that they can seamlessly track their progress while working on different tasks.
However, users also indicate that the tool cannot handle multichannel images: there are occasional lags, the program can run slow during updates, and the UI tends to glitch.
Pricing: LabelBox has a 14-day free trial for small teams and even provides another Pro free trial for companies developing AI models.
6. Playment/ TELUS International
Company history: Playment was founded in 2015 as a managed data labeling platform that generates training data for computer vision models. In 2021, it was acquired by TELUS International, a Canadian technology company that provides IT services and multilingual customer service to global clients.
Data labeling tools: Playment’s Ground Truth (GT) studio is a self-serve data labeling solution that provides ML-assisted 2D and 3D labeling tools for image, video, and sensor fusion annotation.
Services and security: It also prides itself on its extensive feature set, fully-managed labeling services, demonstrated dataset security, built-in quality controls, performance tracking, and powerful APIs for pipeline integration.
G2 user summary: Playment/ TELUS International is the #6 data labeling tool on G2, with a rating of 4.7/5 and 11 reviews. According to most user reviews, the tool stands out in its ability to foster accurate data labeling and management across different sectors to train and validate their model prototypes. Yet, there are some complaints about the pricing being a bit high and the reporting application not being customizable per user and project requirement.
Pricing: Does not have a free plan. No other information is available.
7. Appen
Company history: Founded in 1996, Appen is a licensed platform that allows annotating training data use cases in computer vision and natural language processing. It provides data sourcing, data labeling, and model evaluation. The company’s funding and valuation are private, yet it's one of the oldest solutions in the market and has demonstrated experience in managing data for the AI lifecycle.
Data labeling tools: Appen supports data sourcing (pre-labeled datasets, data collection, synthetic data generation), data preparation, and real-world model-evaluation needs, allowing users to develop and launch models with confidence, saving time to focus on other priorities.
Security: The platform is compliant with data security requirements, especially when dealing with personally identifiable information (PII), protected health information (PHI), and other specific regulations.
G2 user summary: Appen is #7 on G2’s list, with a 4.1/5 rating and 16 reviews. Many users emphasize that Appen’s data labeling process, tracking, and storing are noteworthy. The users explain that the website is simple, easy to use, and provides a wide range of projects. The not-so-positive aspect is that the invoicing method can be challenging, project qualifications can be delayed, servers tend to crash frequently, and many users find themselves flooded with Appen emails.
Pricing: Does not have a free trial.
8. Amazon SageMaker Ground Truth
Company history: Launched in 2018, the Amazon SageMaker Ground Truth was initially built to allow users to identify raw data, add informative labels, and produce labeled synthetic data to create training datasets for machine learning models. It also offers two versions: Amazon SageMaker Ground Truth Plus and Amazon SageMaker Ground Truth.
Data labeling tools: Amazon SageMaker Ground Truth helps users build accurate training datasets for machine learning and AI models in a timely manner.
Project management and services: As a user, here you can not only improve the quality of your training datasets but also set up labeling workflows, apply ML-powered automation, choose your own data labeling workforce, and increase the visibility of data labeling operations.
G2 user summary: The tool is in 8th place on the G2 list with a rating of 4.1/5 and 19 reviews. It was selected as the 3rd easiest to use data labeling software. According to the user reviews, Amazon's product is simple, reduces costs and time, and lessens constant human involvement when it comes to data labeling.
However, many users complain about the pricing being too high and their inability to save both money and storage as the endpoint cannot be turned off.
Pricing: Has a free trial, and for the first two months after using Amazon SageMaker, the user’s first 500 objects labeled per month are free.
9. Innotescus Image and Video Annotation Platform
The data labeling solution does not exist anymore, despite being #9 on G2.
10. Hive
Company history: Founded in 2013, Hive or Hive Data provides cloud-based AI solutions for understanding content and offers turnkey software products powered by proprietary AI models and datasets. Hive has raised a total of $120.7M in funding over 6 rounds.
Data labeling tools and services: Hive's APIs allow engineers to integrate pre-trained AI models that address content understanding needs. Hive's intelligent search APIs power visual similarity and text-to-image search. Here, you can streamline content moderation and labeling, automate search and authentication, and protect digital ownership. Besides, you can monitor and measure cross-platform sponsorship and better monetize premium ad inventory.
G2 user summary: Hive is #10 on the list, with a 4.4/5 rate and 10 reviews. Users find Hive data quite easy to use and effective for labeling and building AI solutions. Some of the downsides include overlapped images, unrecognized data, and slow query performance. There are also mentions of pre-trained models being useless.
Pricing: Does not include any pricing information.
11. Basic AI/Xtreme1
Company history: Established in 2019, Basic AI's Xtreme1 is a one-stop data-centric MLOps platform that ensures data manageability and automation throughout its AI lifecycle.
Data labeling tools: Xtreme1 especially stands out because of its LiDAR data labeling combined with image and video content, which for the most part, serves the autonomous driving industry. In terms of industry-specific tasks, it addresses object and lane detection, object tracking, and semantic segmentation. You can either start with a pre-trained model, integrate an existing one, or continuously train your own model.
G2 user summary: The 11th spot on the list belongs to Basic AI, with a rate of 4.2/5 and 22 reviews. According to most users, BasicAI’s data labeling software is high-quality, easy to use, and can be trained for good results. Users are excited about BasicAI’s support team always being around for help. The setback is that the tool can seem a bit confusing to beginners. Thus, prior knowledge and more training are required. Other areas that need further improvement, as per users, are image detection and image tracking, so they can better fit low-end desktops and laptops.
Pricing: Has a free trial.
12. LinkedAI
Company history: Founded in 2018, LinkedAI is a web platform that allows users to build accurate training datasets using machine learning.
Data labeling tools: LinkedAI provides users with image labeling tools for classification, object detection, and segmentation with automation features.
Project management and services: It offers an end-to-end solution for data annotation with labeling tools, data generation, data management, automation features, and annotation services with integrated tooling.
G2 user summary: LinkedAI is #12, with a rating of 4.6/5 and 20 reviews. Reviewers indicate that LinkedAI’s model development is good, the platform is user-friendly, comprehensible, and provides efficient service for data labeling. On the other hand, many of the users did face issues with PIs monitoring and automated annotation and explained that it needed more training data.
Pricing: LinkedAI is free for students and has a start price of $50 each month per user. The Grow option is $84/mo per user, and the Enterprise plan is customizable.
13. Ango Hub
Company history: Launched in 2020, Ango Hub is an all-in-one data la platform for AI teams. It is often coming off as a data labeling solution for medical AI. Ango Hub has managed to raise $820K throughout its two operating years.
Data labeling tools: Ango Hub is a data annotation platform that provides users with internal tools for data labeling, a real-time issue system, sample label libraries, and more. It offers annotations for images, videos, text, audio, and PDF.
G2 user summary: Ango Hub earned its #13 spot on G2, with a rate of 4.8/5 and 11 reviews.
Users think the platform performs well with heavy video tasks and large PDF files. The tools for labeling text data and classification are quick, and the team is very responsive. However, many users share that there is a learning curve, some features and shortcuts are hard to spot, and the function to zoom into audio waveforms is missing. Users also comment that the UI could be improved, and it is tiring to import assets from the cloud as Ango Hub requires users to make a JSON with individual URLs.
Pricing: Ango Hub has three pricing options: the Free package, which is limited to 5 users and 10k annotations, the Cloud version, and the Enterprise, which will require connecting to the sales team to learn more about the pricing.
14. RedBrick AI
Company history: In 2020, RedBrick AI was established with a workflow system that assists teams in building collaborative and scalable quality assurance processes. As a SaaS platform for labeling workflows, the company’s funding is around $125.0K.
Data labeling tools: RedBrick AI’s platform is used to manage and scale medical annotation needs, including web-based annotations and other training datasets. It can be used to direct, scale, and speed up all medical annotation through a web-based annotation toolset, developer APIs, and much more.
G2 user summary: Redbrick AI is #14 on the list, with a 4.1/5 rating and 8 reviews.
Many users believe RedBrick AI is easy to use and is generally an effective tool for medical annotations. It also provides powerful APIs to the developer. Users also mention that the different features and abilities allow both medical and non-medical companies to use Redbrick AI, which is an advantage.
Some of the drawbacks mentioned include lagging during 3D imaging and the need to make several attempts to get the best result.
Pricing: They might offer a 14-day free trial to a user based on a one-on-one discussion.
15. Scale Rapid
Company history: Founded in 2016, Scale Rapid is a labeling platform for machine learning teams to get training data. Established to solve issues of scaling data labeling pipelines to production-level volumes, the company now has $603M worth of investments.
Data labeling tools: With Scale Rapid, you can label data like 3D sensors, images, and video at speed while maintaining the annotation quality. Except for providing high-quality training data, and precise annotation, Scale Rapid also provides real-time feedback on annotation instructions, accelerating the data labeling process and model development.
Security: Scale cloud platform’s infrastructure and operations are compliant with industry best practice standards and regulations.
G2 user summary: Scale lands the 15th spot on the G2 list, with its 4.4/5 score and 8 reviews.
Users agree on Scale Rapid's ease of use and convenience when it comes to annotating data within a short period of time. However, many do indicate that there is room for improvement and updates, as, sometimes, data gets hard to understand. Users also mention that the company should work on Redbrick AI’s UI and make it more user-interactive. Users would also like to see some price decrease.
Pricing: Does not have a free trial, instead, it offers two packages: Rapid and Enterprise.
16. Supervisely
Company history: Since 2013, the founders have been trying to build end-to-end solutions for clients from different parts of the world. Yet it was not until 2017 that Supervisely saw the light of day when the company changed its primary focus from services to products.
Data labeling tools and project & QA management: Supervisely offers image, video, DICOM, and LIDAR labeling. Here you can additionally manage datasets, perform quality assurance on your data and train high-performance neural networks.
G2 user summary: The #16 on the list is Supervisely, with a score of 4.8/5 and 9 reviews. Users observe significant performance improvement while using Supervisely. The solution enables them to establish a platform that can integrate a large number of open-source tools and custom-built solutions.
Yet users also share that the UI can be tricky and overwhelming for new users and that the platform speed could be improved.
Pricing: Offers a free Community version of the tool while also providing a 30-day free trial for the Enterprise version.
17. TrainingData.io
Company history: Initially established in 2018, TrainingData.io’s founding team has over 20 years of combined experience in building robust solutions for Visual AI. Training Data.io offers image and video training dataset labeling tools for radiology, pathology, and many other forms of medical data.
Data labeling tools and security: TrainingData.io provides pixel annotation tools, annotator performance management, labeling instruction builder, and data security. It offers high-precision training data labeling and uses AI-assisted features to speed up the work of machine learning engineers.
G2 user summary: With a score of 4/5 and 7 reviews, TrainingData.io is #17 on the G2 list. In these reviews, the users state that the software is easy to use — especially image and video data labeling — and the customer support team is very helpful. Yet many complain about the time-consuming aspect of training. Some even assure that with large datasets, the system may even crash. There are also mentions of the website being down most of the time and the pricing being too high compared to other similar software.
Pricing: TrainingData.io offers 4 different payment plans: a free version, Pro, Radiology, and Enterprise, and each differs in price.
18. Kili
Company history: What started as a simple business idea in 2018 is now known as Kili. The goal of the two founders was to ensure that data is no longer a barrier to good AI. By 2020, the Kili platform went live, began operating as a data labeling tool, and has managed to raise a total of $31.9M in funding.
Data labeling tools: Kili's AI training data platform assists large organizations in transitioning from "big data" to "good data."
DataOps and services: The platform combines collaborative data (image, video, text, audio, and OCR) annotation with data-centric workflows, automation, curation, integration, and simplified DataOps to create high-quality AI. Moreover, Kili offers a fully managed expert labeling workforce to seamlessly ramp up projects without having in-house annotators on board.
G2 user summary: Kili is #18 with a rating of 4.3/5 and 6 reviews. Kili is proved useful to users because of its ability to collaborate among different teams. However, some, again, indicate that the tool cannot handle massive loads of data during the training phase. The project creation process can be time-consuming and tiresome.
Pricing: Kili has three different price packages: a free Community version, a Start Custom Plan, and an Enterprise Plan.
19. Shaip
Company history: the idea of creating Shaip was initiated in 2018 when the two founders met a Fortune 10 company client. Their initial goal was to organize medical data to enhance patient care and decrease the costs of healthcare. Now, Shaip is a fully managed data platform that addresses the most pressing AI challenges.
Data labeling tools: The Shaip cloud platform is designed to label images, videos, text, speech, and audio, empowering more teams to build AI products. It is a human-in-the-loop ML platform that also offers specialty solutions grouped by industries.
G2 user summary: With a 4.1/5 rating and 5 reviews, the #19 on the G2 list is Shaip Cloud. The reviews indicate that Shaip Cloud is a solid human-in-the-loop ML platform that helps label and manage training datasets for chatbots and NLP. On the other hand, many reviews confirm that instruction and training may be required in advance to be able to take full advantage of the tool. Many also mention that the company needs to work on identification and speech recognition tasks. There were also mentions of the software failing to provide meaningful results.
Pricing: Does not offer a free trial, and the website does not provide any pricing information.
20. Super.ai
Company history: Long before the establishment of Super.ai, the founder, Brad, had created TrueMotion. It was not until 2017 that Super.ai started to provide the same technology used to build and utilize machine learning algorithms at TrueMotion. At this point of writing, Super.ai is estimated to have $18.3M in funding.
Data labeling tools: Super.ai is used to structure and label any type of data and automate the processing of images, videos, text, and audio. The platform's key capabilities also address data integration, AI workflows, quality controls, an active trainer, and a smart combiner (combining results from multiple annotators into a single output).
G2 user summary: Super.ai is #20 on G2, with a score of 4.5/5 and 5 reviews. When listing the best things about Super.ai, the users recall automated workflows and the ease of turning unstructured data into AI applications. Yet, many also complain about feature limitations and the cost of the platform.
Pricing: Offers a wide range of free trials.
21. Sama
Company history: Founded in 2008, Sama was built with the goal of advancing AI development through an accurate, scalable, and ethical data pipeline. Sama has raised $70M of Series B funding to devise the very first end-to-end AI platform. Ever since, the company has been providing users with a platform that can support their models from end to end.
Data labeling tools: Sama specializes in image, video, 3D point cloud, sensor data labeling, validation for machine learning algorithms, and data curation.
G2 user summary: Sama is #21 on G2, with a 4.5/5 rating and 5 reviews. According to users, it provides proper annotation, delivers data at the right time, and has the ability to break down complex data into small tasks. Yet, many acknowledge the need for ML model training as a part of the solution and cost reduction.
Pricing: Has a free trial for self-service tools.
22. KeyLabs
Company history: Initially founded as an image and video data labeling service provider in 2019, KeyLabs wanted to expand and build its own toolset, gradually transforming into the all-in-one annotation platform it is now. Their funding is around $1.2M at this point of writing.
Data labeling tools and project management: Keylabs is a labeling solution for images and videos that provides users with a palette of annotation types, automating the process of building, deploying, and maintaining machine learning models. In addition, Keylabs provides space for annotation project management, quality assurance, and dataset verification.
G2 user summary: KeyLabs is #22 on the list, with its 4.8/5 score and 3 reviews. Users believe KeyLabs is user-friendly, with a limitless capacity for video labeling. However, the tool lacks LIDAR annotation and natural language processing.
Pricing: KeyLabs offers a free trial and has 4 different pricing options: Startup, Business, Pro, and Enterprise.
23. Clarifai
Company history: Matthew Zeiler developed Clarifai based on his 2013 ImageNet win. Later in 2019, the company was named a leader in Forrester’s New Wave Computer Vision Platforms report, being the only startup to receive a differentiated rating. Clarifai offers a solution for building AI-powered software and has raised a total amount of $100M.
Data labeling tools and services: Clarifai is an end-to-end AI platform for computer vision, natural language processing, and audio recognition. It transforms unstructured images, videos, texts, and audio data into AI-useful data. Other important features of Clarifai involve AI workflows, data labeling services, custom model building, etc.
G2 user summary: Labeled as G2’s Easiest Admin of Fall 2022, Clarifai is #23 on the list, with a 4.3/5 rating and 17 reviews. Users comment on the simplicity of the tool and especially love that it does not require prior programming knowledge. There are also mentions of how the APIs are available in a structured and systematic way. Though users also have concerns about customer service, the lack of online support forums, and optimization on the NLP side.
Pricing: Clarifai offers three versions for its users, a free Community version, an Essential version which starts at $30/mo, and a Professional one which starts at $300/mo.
24. Diffgram Data Training Software
Company history: Founded in 2018, Diffgram is an open-source training data platform that aims to provide its users with an unlimited experience. Their total funding amount is $1.5M.
Data labeling tools and project management: Diffgram is an open-source data labeling software where you can annotate the following data formats: 3D, image, video, audio, text, geospatial, and DICOM. The other two major components of the platform are the workflow and the catalog, where you can explore, visualize and manage your unstructured data (curation).
G2 user summary: The #24 on G2's list is the Diffgram Data Training Software, with a 5/5 rate and 2 reviews. One user commented that it provides the ability to annotate unlimited data without having to worry about the billing, as it is open-source. Another user mentioned that it is sensitive to rough edges.
Pricing: It offers two options; the open-source, free version and the Enterprise, which requires further communication with the team to learn about the prices.
25. Heartex
Company history: Established in 2019, Heartex is a labeling management system that powers internal data labeling operations to build the most competitive ML/AI models at scale. Over the past few years, the company has raised $30M in funding.
Data labeling tools, project management, and security: Heartex's platform consists of three major categories: collaborative labeling suite (roles and permissions, workspaces, and label queue management), QA and analytics (review workflows, annotator agreement matrix, management reports), and secure cloud-based service (data stays on your local servers, SSO & LDAP Integrations, and SOC2 certified).
G2 user summary: With a score of 3.5/5 and 2 reviews, Heartex is #25 on G2.
Those two reviews mention that the tool minimizes the amount of time spent on labeling and that it has a good security system in place. However, there is an issue with the pricing, as it tends to be on the higher side. Another user experienced challenges when connecting with an SSO provider.
Pricing: Heartex offers two pricing options; a Community version that is free and open-source and an Enterprise one that has custom pricing.
26. Plainsight
Company history: Funded in 2007, Plainsight streamlines vision AI for enterprises and offers them the ability to build, manage, and operationalize solutions. Till now, the company has raised around $33.9M.
Data labeling tools and project management: Plainsight is a data annotation tool for computer vision that additionally offers AI-powered automation, model, dataset, and pipeline process versioning, deployment, as well as maintenance, and long-term oversight to prevent bias and revenue loss.
G2 user summary: Plainsight is #26, with its 4.3/5 rating and 3 reviews. In these three reviews, users agree on the simplicity of the tool and feature variety. What users found a drawback was the technical knowledge requirement and limits for imported and exported files.
Pricing: Plainsight offers a free trial for the on-demand version and has another Enterprise option where the pricing is based on business needs.
27. Alegion
Company history: Founded in 2012, Alegion is annotation software that creates reliable training data for machine learning models, combining human-in-the-loop data labeling with an AI-enabled platform. At the moment of writing, the company’s funding is around $33.2M.
Data labeling tools: Alegion addresses fundamental AI and training data pain points across industries, enabling image, video, and NLP annotation.
Project management and security: Alegion also has a self-serve solution with comprehensive project management as well as a managed services marketplace.
G2 user summary: Alegion is #27, with a single five-start review praising its labeling accuracy.
Pricing: Has 3 pricing options; a Control (self-serve) solution where the first 150 hours are free, a Managed Platform, and Managed Labeling Service, which requires contacting an Alegion expert for pricing details.
28. DefinedCrowd
Company history: Defined.ai positions itself as one of the world’s leading online marketplaces, where AI professionals are able to commission, buy, or sell AI training data, tools, or models. As of 2022, Defined.ai has raised a total of $78.6M in funding over 6 rounds.
Data labeling tools and services: On DefinedCrowd, you’ll find a professional AI marketplace to help you build top-notch AI. You can also monetize your AI data, tools, or models by applying to become a vendor for speech, NLP, computer vision, and machine translation annotations.
G2 user summary: DefinedCrowd is the 28th one on the list, again with a single five-star review.
The review mentions that "although the pay was higher than other crowd worker platforms, DefinedCrowd is a great crowdsourcing platform for those looking to make money working from home."
Pricing: No pricing information on the website.
29. Hasty
Company history: Before launching Hasty in 2019, the core team worked on asset tracking, robotics, sorting, and QA applications with various German manufacturers using vision AI. The company was built around the idea of utilizing AI to train AI. Hasty.ai has raised a total of $3.7M in funding over a single round.
Data labeling tools and security: Hasty.AI offers an end-to-end unified vision AI platform that improves data labeling and reduces the cost of data quality control. The top product components are data labeling, model playground, quality control, and data security (HIPAA and ISO27001-certified, available on-premise), which pretty much encompasses everything needed to build AI applications.
G2 user summary: Hasty.Ai is #29 on G2 with a single 4-star review, mentioning that Hasty.AI’s user interface is impressive and easy to use. The cons are that the user needs to customize UIs and have security options for specific organizations.
Pricing: Has 3 different options: a free starter package, a self-serve plan, and a custom one.
30. Labeling AI
This labeling tool is not active anymore (a single 4-start review).
Got tons of data and don’t know where to start? Join 100+ companies supercharging their annotation pipelines with SuperAnnotate.
