Akiwi

Akiwi.eu makes keywording any image a breeze, letting anyone quickly generate spot-on keywords with just a simple drag-and-drop.

Visit akiwi.eu

Akiwi.eu is a user-friendly tool that simplifies the process of generating accurate and relevant keywords for any uploaded image. By using a straightforward drag-and-drop interface, it allows users to quickly refine and select keywords, making image tagging more efficient and effective.

Facts about Akiwi:

22 million reference images
15 seconds to generate keywords
15+ keywords per image

Key Features

Support for Common Image Formats: Akiwi seamlessly handles all major browser-supported image formats, including JFIF, PJPEG, JPEG, PJP, JPG, and PNG, ensuring compatibility with your images.

Tagging of Private Images: Akiwi allows you to securely keyword your private images, with no storage of your photos on the platform, ensuring your content remains private.

Web Application, No Installation Needed: Akiwi is a fully web-based tool that requires no installation, allowing you to access and use it directly from your browser, anytime and anywhere.

How to Use Akiwi?

Drag and drop your image onto the Akiwi platform.
Akiwi will analyze the image and display similar reference images from its extensive database.
Select the image that most closely matches your own to refine the keyword suggestions.
Akiwi will generate a list of relevant keywords based on your selection.
Refine the list further by selecting or deselecting keywords as needed.
Once satisfied, copy the final set of keywords to your clipboard.

How Does Akiwi Work?

Akiwi.eu is a sophisticated keywording tool that leverages advanced machine learning techniques to analyze and tag images efficiently. At its core, the system processes the visual content of images to generate high-dimensional feature vectors, which capture the essential characteristics of the image. These vectors are then compressed and quantized using custom-developed neural compression networks, reducing them to 64-dimensional representations, known as "fingerprints." These fingerprints serve as a unique identifier for each image, enabling the system to quickly and accurately find other images with similar content. By comparing these fingerprints, Akiwi efficiently suggests relevant keywords, simplifying the tagging process for photographers and contributors.

Watch the video on YouTube

Best practice when using Akiwi

Publications

List of relevant publications

Deep Metric Learning using Similarities from Nonlinear Rank Approximations

Konstantin Schall, Kai Uwe Barthel, Nico Hezel, Klaus Jung
2019 IEEE 21st International Workshop on Multimedia Signal Processing (MMSP '19)

We introduce a metric learning algorithm that enhances image similarity search by focusing on feature vectors that most impact retrieval quality. By computing normalized approximated ranks and using a nonlinear transfer function in a new loss function, our approach significantly improves deep feature embeddings across multiple datasets.

Deep Aggregation of Regional Convolutional Activations for Content Based Image Retrieval

Konstantin Schall, Kai Uwe Barthel, Nico Hezel, Klaus Jung
2019 IEEE 21st International Workshop on Multimedia Signal Processing (MMSP '19)

We present a supervised aggregation method for deep learning-based image retrieval that combines regional pooling with weighted activation averages to create a highly representative feature vector. Our approach, which includes fine-tuning with a new NRA loss function, achieves state-of-the-art results on the INRIA Holidays dataset and competitive results on the Oxford Buildings and Paris datasets, while significantly reducing training time.

List all publications