At the turn of this decade, there is a surge of no-code AI platforms. More and more businesses are looking to leverage the power of artificial intelligence to build smarter software-based products.

在這十年之交，無程式碼AI平臺激增。 越來越多的企業正在尋求利用人工智慧的力量來構建更智慧的基於軟體的產品。

But execution becomes an obstacle for many. It’s a challenge for startups to find people with relevant machine learning expertise as the field is always a work in progress.

但是執行成為許多人的障礙。 對於初創公司而言，尋找具有相關機器學習專業知識的人才是一項挑戰，因為該領域一直處於發展之中。

A lot of firms that invest fortunes in hiring engineers with PhDs and academic research background in machine learning fail to launch their products.

許多在招聘具有博士學位和具有機器學習學術研究背景的工程師方面投入大量資金的公司未能推出他們的產品。

This brings no-code visual drag-and-drop tools to the picture which helps fill the data scientist’s void and makes artificial intelligence less intimidating for the non-technical people.

這為圖片帶來了無需程式碼的視覺化拖放工具，這有助於填補資料科學家的空白，並減少了人工智慧對非技術人員的威脅。

Businesses can now generate datasets, train, and deploy models with minimal to no coding knowledge in significantly less time while staying economical.

企業現在可以在花費最少的時間內，以最少的編碼知識甚至沒有編碼知識就可以生成資料集，訓練和部署模型，同時保持經濟性。

For mobile application developers, this certainly is a boon in disguise as on-device machine learning is in high demand right now. They don’t need to have a Ph.D. in machine learning and can be more creative with the data and models they wish to train.

對於移動應用程式開發人員而言，這無疑是一種變相，因為目前對裝置上機器學習的需求很高。 他們不需要博士學位。 機器學習方面的知識，可以利用他們希望訓練的資料和模型更具創造力。

In the next few sections, we’ll walk through some of the best no-code machine learning tools available right now. Some of these are totally free while others might charge you beyond their free trials. Nevertheless, each of them will help you to bring your AI application ideas to reality.

在接下來的幾節中，我們將逐步介紹一些目前可用的最佳無程式碼機器學習工具。 其中一些是完全免費的，而其他一些可能會向您收取超出免費試用期的費用。 但是，它們每個都將幫助您將您的AI應用程式想法變為現實。

建立ML (Create ML)

Being an iOS developer, I had to start with Apple’s no-code drag and drop tool, CreateML. After initially launching with Xcode, today CreateML is an independent macOS application that comes with a bunch of pre-trained model templates.

作為iOS開發人員，我必須從Apple的無程式碼拖放工具CreateML開始。 在最初使用Xcode啟動之後，今天的CreateML是一個獨立的macOS應用程式，帶有一堆預先訓練的模型模板。

By using transfer learning lets you build your own custom models. From image classifiers to style transfers to natural language processing to recommendation systems it has almost every suite covered. All you need to do is pass the training and validation data in the required formats.

通過使用轉移學習，您可以構建自己的自定義模型。 從影象分類器到樣式轉換，再到自然語言處理再到推薦系統，它幾乎涵蓋了所有套件。 您所需要做的就是以所需格式傳遞培訓和驗證資料。

Moreover, you can fine-tune the metrics and set your own iteration count before starting the training. Create ML provides realtime results on the validation data for models such as style transfer. In the end, it’ll generate a CoreML model that you can test and deploy in your iOS applications.

此外，您可以在開始訓練之前微調指標並設定自己的迭代計數。 Create ML在諸如樣式轉換之類的模型的驗證資料上提供實時結果。 最後，它將生成一個CoreML模型，您可以在iOS應用程式中對其進行測試和部署。

Google AutoML(Google AutoML)

While Apple is leading the way with Create ML, Google couldn’t afford to be left behind. There AutoML tool works much the same way as CreateML albeit on the cloud.

儘管蘋果公司在Create ML方面處於領先地位，但Google卻不甘落後。 儘管在雲上，但AutoML工具的工作方式與CreateML幾乎相同。

Google’s Cloud AutoML currently includes Vision(image classification), Natural Language, AutoML Translation, Video Intelligence, Tables in its suite of machine learning products.

Google的Cloud AutoML當前在其機器學習產品套件中包括視覺(影象分類)，自然語言，AutoML翻譯，視訊智慧，表格。

This enables developers with limited machine learning expertise to train models specific to their use cases. AutoML on the cloud removes the need to know transfer learning or how to create a neural network by providing out of the box support for thoroughly tested deep learning models.

這使具有有限機器學習專門知識的開發人員可以訓練針對其用例的模型。 通過為經過全面測試的深度學習模型提供開箱即用的支援，雲上的AutoML無需瞭解遷移學習或如何建立神經網路。

Once the model training is finished you can test and export the model in .pb ,.tflite , CoreML etc formats.

模型訓練完成後，您可以測試並匯出.pb ， .tflite ，CoreML等格式的模型。

MakeML(MakeML)

MakeML is a developer tool used for creating object detection and semantic segmentation models without code.

MakeML是一個開發人員工具，用於建立無需程式碼的物件檢測和語義分段模型。

It provides a macOS app for iOS developers to create and manage datasets(such as performing object annotations in images). Interestingly, they also have a dataset-store with some free computer vision datasets to train a neural network in just a few clicks.

它為iOS開發人員提供了macOS應用程式，用於建立和管理資料集(例如在影象中執行物件註釋)。 有趣的是，他們還擁有一個數據集儲存區，其中包含一些免費的計算機視覺資料集，只需單擊幾下即可訓練神經網路。

MakeML have shown their potential in sports-based applications wherein you could do ball tracking. Also, they have an end to end tutorials for training nail and potato segmentation models which should give any non-machine learning developer a good headstart.

MakeML在基於運動的應用程式中展示了其潛力，您可以在其中進行球追蹤。 此外，他們還有用於培訓指甲和土豆細分模型的端到端教程，這些教程應為任何非機器學習開發人員提供良好的起點。

Using their built-in annotation tool that works in videos you can build a hawkeye detector that’s used in cricket and tennis games.

使用它們在視訊中使用的內建註釋工具，您可以構建用於板球和網球比賽的鷹眼檢測器。

弗裡茨·艾裡(Fritz AI)

Fritz AI is a growing machine learning platform that helps bridge the gap between mobile developers and data scientists.

Fritz AI是一個不斷髮展的機器學習平臺，可幫助縮小移動開發人員和資料科學家之間的鴻溝。

iOS and Android developers can quickly train and deploy models or use their pre-trained SDK which provides out of the box support for style transfer, image segmentation, pose estimation like models.

iOS和Android開發人員可以快速訓練和部署模型，或者使用他們的預先訓練的SDK，該SDK為樣式轉移，影象分割，姿勢估計等模型提供現成的支援。

Their Fritz AI Studio lets you quickly turn ideas into production-ready apps by providing data annotation tools and synthetic data to generate datasets in a seamless fashion.

他們的Fritz AI Studio通過提供資料註釋工具和合成資料以無縫方式生成資料集，使您能夠快速將創意轉變為可用於生產的應用程式。

Besides introducing support for Style Transfer before Apple, Fritz AI’s machine learning platform also provides solutions for model retraining, analytics, easy deployment, and protection from attackers.

除了在Apple之前推出對Style Transfer的支援外，Fritz AI的機器學習平臺還提供了模型重新訓練，分析，易於部署以及免受攻擊者保護的解決方案。

跑道ML(RunwayML)

Here’s another great machine learning platform designed specifically for creators and makers. It provides a delightful visual interface to quickly train models ranging from text and image generation(GANs) to motion capture, object detection, etc without the need to write or think in code.

這是專門為創作者和製作者設計的另一個出色的機器學習平臺。 它提供了令人愉悅的視覺介面，可快速訓練從文字和影象生成(GAN)到運動捕捉，物件檢測等各種模型，而無需編寫或思考程式碼。

RunwayML lets you browse a range of models ranging from super-resolution images to background removal and style transfer.

通過RunwayML ，您可以瀏覽一系列模型，從超解析度影象到背景去除和樣式轉移。

While exporting models from their application isn’t free of cost, a designer can always leverage the power of their pre-trained generative adversarial networks to synthesize new images from their prototypes.

儘管從其應用程式中匯出模型並非沒有成本，但設計人員始終可以利用其預先訓練的生成對抗網路的功能來從其原型中合成新影象。

Their Generative Engine that synthesizes images as you type sentences is one of the highlights. You can download their application on macOS, windows or use it on the browser directly(currently in beta).

它們的生成引擎可以在您鍵入句子時合成影象，這是其中的亮點之一。 您可以在macOS，Windows上下載其應用程式，也可以直接在瀏覽器中使用它們(當前為Beta)。

顯然是AI(Obviously AI)

Obviously AI uses state of the art natural language processing to perform complex tasks on user-defined CSV data. The idea is to upload the dataset, pick the prediction column, and enter questions in natural language and evaluate results.

顯然，AI使用最先進的自然語言處理來對使用者定義的CSV資料執行復雜的任務。 想法是上傳資料集，選擇預測列，然後以自然語言輸入問題並評估結果。

The platform trains the machine learning model by choosing the right algorithm for you. So, just with a few clicks, you can get a prediction report be it for forecast revenue or predicting the inventory demand. This is incredibly useful for small and medium-sized businesses looking to get a foot into the field of artificial intelligence without having an in-house data science team.

該平臺通過為您選擇合適的演算法來訓練機器學習模型。 因此，只需單擊幾下，您便可以獲得預測報告，無論是用於預測收入還是預測庫存需求。 這對於希望在沒有內部資料科學團隊的情況下涉足人工智慧領域的中小型企業非常有用。

Obviously AI lets you integrate data from other sources as well such as MySQL, Salesforce, RedShift, etc. So, without knowing how linear regression and text classification you can leverage their platform to run predictive analysis on your data.

顯然，AI使您可以整合來自其他來源(例如MySQL，Salesforce，RedShift等)的資料。因此，在不知道線性迴歸和文字分類如何的情況下，您可以利用其平臺對資料進行預測分析。

超級註釋(SuperAnnotate)

Beyond model training, data processing eats up a major chunk of time in developing machine learning projects. Cleaning and labeling data can certainly consume lots of hours especially when you’re dealing with thousands of images.

除了模型訓練之外，資料處理還佔用了開發機器學習專案的大部分時間。 清理和標記資料肯定會耗費大量時間，尤其是在處理成千上萬張影象時。

SuperAnnotate is an AI-powered annotation platform that uses machine learning capabilities(specifically transfer learning) to boost your data annotation process. By using their image and video annotation tools you can quickly annotate data with the help of built-in predictive models.

SuperAnnotate是一個由AI驅動的註釋平臺，它使用機器學習功能(特別是轉移學習)來增強您的資料註釋過程。 通過使用他們的影象和視訊註釋工具，您可以在內建的預測模型的幫助下快速註釋資料。

So, generating datasets for object detection, image segmentation will get a whole lot easier and faster. SuperAnnotate also handles duplicate data annotation which is common in video frames.

因此，生成用於物件檢測的資料集，影象分割將變得更加容易和快捷。 SuperAnnotate還可以處理視訊幀中常見的重複資料註釋。

教學機(Teachable Machine)

Last but not the least, we have another Google no-code machine learning platform. Unlike, AutoML which is a little developer-friendly, Teachable Machines let you quickly train models to recognize images, sounds, and poses right from your browser.

最後但並非最不重要的一點是，我們還有另一個Google無程式碼機器學習平臺。 與AutoML不同，它對開發人員有點友好，可教學機器使您可以快速訓練模型以從瀏覽器直接識別影象，聲音和姿勢。

You can simply drag and drop files to teach your model or use the webcam to create a quick and dirty dataset of images or sounds. Teachable Machine uses the Tensorflow.js library in your browser and ensures that your training data stays on the device.

您可以簡單地拖放檔案來教您的模型，也可以使用網路攝像頭來建立影象和聲音的快速而骯髒的資料集。 Teachable Machine使用瀏覽器中的Tensorflow.js庫，並確保您的訓練資料保留在裝置上。

This is certainly a big step by Google for people who wanted to practice machine learning without any coding knowledge. The final model can be exported in Tensorflow.js or tflite formats which can then be used in your websites or app. You can also convert the model into different formats using Onyx.

對於那些想在沒有任何編碼知識的情況下進行機器學習的人們來說，這無疑是Google邁出的一大步。 最終模型可以Tensorflow.js或tflite格式匯出，然後可以在您的網站或應用中使用。 您也可以使用Onyx將模型轉換為不同的格式。

Here’s a simple image classification model I managed to train in less than a minute.

這是我在不到一分鐘的時間內便完成的簡單影象分類模型。

結論(Conclusion)

We saw how no code machine learning platforms bridge the gap between data scientists and non-ML practitioners. While there’s no one size fits all solution, you can always pick a platform to build models or generate datasets at express speed.

我們看到了沒有程式碼機器學習平臺如何彌合數據科學家和非ML實踐者之間的鴻溝。 儘管沒有一個適合所有解決方案的規模，但是您始終可以選擇一個平臺來快速構建模型或生成資料集。

Moreover, such tools make machine learning a lot more fun to work with. SnapML is another great no code machine learning tool that lets you train or upload your own custom models and use in Snap Lenses. This certainly helps indie developers and creators to put forth their creativity in front of millions of people.

而且，這樣的工具使機器學習變得更加有趣。 SnapML是另一個很棒的無程式碼機器學習工具，可讓您訓練或上傳自己的自定義模型並在Snap Lenses中使用。 這無疑有助於獨立開發人員和創作者在數百萬人面前展現他們的創造力。

That’s it for this one. Thanks for reading.

就這個。 謝謝閱讀。