Example Dataset (Tracking SAM2)

## Overview

Example Dataset (Tracking SAM2) is a dataset for instance segmentation tasks - it contains Fish annotations for 450 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Wydruki 3D

## Overview

Wydruki 3D is a dataset for instance segmentation tasks - it contains 3d Prints annotations for 550 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Here are a few use cases for this project:

1. Online Board Game Platforms: The model can be used to digitize physical board games. For example, a user can play a chess game on a physical board, then use computer vision to transfer the game's state into a digital format.

2. Board Game Analysis and Strategy Development: Chess coaches or enthusiasts can use it to analyse games. They can take pictures or real-time video of their games, and the AI can determine the board configuration, facilitating the analysis of the moves.

3. Educational Tools: The model could be used in educational applications, providing a digitized board setting for learning purposes. For example, teaching different types of chess strategies and tactics.

4. Search Engines and E-commerce: The model could help improve search results or product categorization by identifying different types of boards in images.

5. Surveillance Systems: In a retail setting, the model can detect when board games or similar merchandise are removed from a shelf, enhancing loss prevention.

This project is a product of the Hodin Lab at the University of Washington Friday Harbor Laboratories. It has the goal of developing a Pacific Northwest sea star instance segmentation and classification tool. Ultimately this tool will be used as a part of a sea star photo re-identification pipeline. Additionally, we hope this tool aids in the use of camera transect surveys of marine habitat.

The primary target species for this model is the Sunflower Seastar Pycnopodia helianthoides, the inclusion of other species is to make the model more robust to confusion species when deployed. For this reason, and our labs access to images of the Sunflower seastar it is over represented in the dataset.

If you have images of sea stars and wish to contribute to the project contact Willem @ willemlw@uw.edu

Our target number of annotated images is >10k with >100 annotated examples for each species.

We hope to have a future extension of this model which includes both star and prey annotations.

We are drawing images from a diverse set of sources including. iNaturalist Google search Collaborators Personal lab + field images boldsystems

We are using the annotation tool CVAT

Involved members are: Willem Lee Weertman Marilyn Duncan Ian Taylor Jason Hodin Brook Ashcraft

Open biomass burning (OBB) significantly impacts regional and global air quality, climate change, and human health. It is susceptible to fire types, including forests, shrublands, grasslands, peatlands, and croplands burning. Global high-resolution satellites have advantages in detecting active fires, enabling more accurate estimation of these emissions. In this study, we develop a global high-resolution (1 km×1 km) daily emission inventory associated with OBB emissions using the Chinese Fengyun-3D satellite’s global fire spot monitoring data, satellite and observational biomass data, vegetation index-derived spatiotemporal variable combustion efficiency, and land type-based emission factors. The results showed that the average annual OBB emissions for 2020–2022 were 2,586.88 Tg C, 8841.45 Tg CO2, 382.96 Tg CO, 15.83 Tg CH4, 18.42 Tg NOX, 4.07 Tg SO2, 18.68 Tg OC, 3.77 Tg BC, 5.24 Tg NH3, 15.85 Tg NO2, 42.46 Tg PM2.5 and 56.03 Tg PM10. More specifically, taking carbon emissions as an example, the average annual OBB for 2020–2022 were 72.71 (BONA), 165.7 (TENA), 34.1 (CEAM), 42.9 (NHSA), 520.5 (Southern Hemisphere South America; SHSA), 13 (EURO), 8.4 (MIDE), 394.3 (Northern Hemisphere Africa; NHAF), 847 (Southern Hemisphere Africa; SHAF), 167.4 (BOAS), 27.9 (CEAS), 197.3 (Southeast Asia; SEAS), 13.2 (EQAS), and 82.4 (AUST) Tg. SHAF was identified as the regions with the largest emissions. Notably, Savanna Grassland accounted for the lion's share of the total emissions, contributing a substantial 46%, followed by Woody Savanna/Shrubs at 33%. Moreover, a notable seasonal variability characterizes OBB carbon emissions, with a marked escalation observed in July and August. This surge in carbon emissions is chiefly attributed to fires in Savanna Grasslands, Woody Savanna/Shrubs, and Tropical Forests of SHAF, SHSA, and NHAF. Fires in Savanna Grasslands were predominant in NHAF, contributing to 77% of emissions during January–April, while in SEAS, Woody Savanna/Shrubs (52%) and Tropical Forests (23%) were the primary sources. Our comprehensive high–resolution inventory of OBB emissions provide valuable information for enhancing the accuracy of air quality modelling, atmospheric transport and biogeochemical cycle studies.

This is an Instance Segmentation project for visualizing detected cracks on concrete. This dataset is usable for those doing transportation and public safety studies, creating self-driving car models, or testing out computer vision models for fun. * Featured Transportation Projects

Check out this example guide from Augmented Startups to see the model in action and learn how the dataset came together: https://medium.com/augmented-startups/yolov7-segmentation-on-crack-using-roboflow-dataset-f13ae81b9958

https://miro.medium.com/max/720/1*BvZk2Sck6cucZ416zgbzMg.jpeg" alt="Inferred Image from YOLOv7">

Here are a few use cases for this project:

1. Retail Inventory Management: Fast_mask can be used by e-commerce and brick-and-mortar stores for identifying and categorizing different product images like jeans, shirts, dresses, and jackets. With its help, retailers can automate the process of inventory management.

2. Automated Fashion Suggestions: Online fashion platforms can use the model to recommend similar pieces of clothing to customers based on the input image. For example, suggesting different styles of blue jeans based on the input pair.

3. Quality Control in Clothing Manufacturing: Businesses involved in manufacturing clothes can use the model to identify any irregularities or defects in their products, such as an unusual color of blue on jeans, ensuring that only quality products reach the market.

4. Clothing Recycling: In waste management and recycling systems, fast_mask could be employed to identify and sort different types of clothing materials for the purpose of recycling.

5. Custom Clothing Developments: Fast_mask could be employed in designing software to identify specific clothing styles, colors, and trends. Based on an input image, the software could provide suggestions to the user about what clothing to design next.

Eggplant

## Overview

Eggplant is a dataset for instance segmentation tasks - it contains Leaves annotations for 504 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [Public Domain license](https://creativecommons.org/licenses/Public Domain).

Correction of the dataset includes:

1. Correct misclassified annotations
2. Annotate missed detections
3. Re-draw the bounding boxes of wrinkles if they are not upto the mark
4. Delete bounding boxes if they do not contain the object
5. For lines on the upper eyelid put them in class "2"

TACO: Trash Annotations in Context Dataset

From: Pedro F. Proença; Pedro Simões

• For more information, go to: http://tacodataset.org
• https://github.com/pedropro/TACO https://raw.githubusercontent.com/wiki/pedropro/TACO/images/logonav.png" alt="TACO Logo">

TACO is a growing image dataset of trash in the wild. It contains segmented images of litter taken under diverse environments: woods, roads and beaches. These images are manually labeled according to an hierarchical taxonomy to train and evaluate object detection algorithms. Annotations are provided in a similar format to COCO dataset.

The model in action:

https://raw.githubusercontent.com/wiki/pedropro/TACO/images/teaser.gif" alt="Gif of the model running inference">

Examples images from the dataset:

https://raw.githubusercontent.com/wiki/pedropro/TACO/images/2.png" alt="Example Image #2 from the Dataset"> https://raw.githubusercontent.com/wiki/pedropro/TACO/images/5.png" alt="Example Image #5 from the Dataset">

For more details and to cite the authors:

• Paper: https://arxiv.org/abs/2003.06975
• Paper Citation: @article{taco2020, title={TACO: Trash Annotations in Context for Litter Detection}, author={Pedro F Proença and Pedro Simões}, journal={arXiv preprint arXiv:2003.06975}, year=

Here are a few use cases for this project:

1. E-Commerce Recommendation System: The "shoe_unsplash" model could be used to create a shoe recommendation system for an e-commerce platform. Based on identified attributes from a customer's previous purchases or viewed products, the system could recommend similar shoes.

2. Shoe Manufacturing Quality Control: The model can be used to detect certain attributes in the production phase. For example, if a shoe was supposed to have a double stitch but the model identifies it doesn't, this could trigger a quality control check.

3. Shoe Sorting in Retail Stores: With camera systems, the model can help automate sorting of shoes based on different attributes. For instance, storing all shoes with laces, short heels, or double stitches in designated areas.

4. Interactive Shopping Experience in Physical Stores: Customers could use an app powered by the "shoe_unsplash" model to identify certain shoe attributes they're looking for while shopping in a physical store. The app would then guide them to the shoes with the desired attributes.

5. Training AI for Video Game Character Design: Game developers could use the model to train AI systems that generate customizable video game characters. It would aid in including diverse options for character footwear, enhancing the gaming experience.

ABESIT Logo

## Overview

ABESIT Logo is a dataset for instance segmentation tasks - it contains Logo annotations for 202 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

In the context of this project, the samples for µ-FTIR analysis contained up to a few thousands particles. The integrated particle detection tool (Particle Wizard - OMNIC Picta) gave poor performances and an AI segmentation tool was needed. Using this dataset, we trained a Detectron2 neural network that was used within GEPARD, an open source software used to improve Raman and FTIR target acquisition and data analysis. With Roboflow, it is possible to export this dataset to various format and use these data to train different architecture of segmentation neural networks.

Here are a few use cases for this project:

1. Quality Control in Wafer Manufacturing: The AI model can be used in wafer manufacturing facilities to automatically classify wafers and identify any flaws or irregularities like "quebrado" (broken), "ratado" (bitten), or "soltando_tampo" (losing top). This could significantly reduce the amount of time required for manual inspection and improve overall production efficiency.

2. Semiconductor Industry Inspection: The "waffer" model can aid in the semiconductor industry to classify and analyze silicon wafers in real-time during the production process. This could help in maintaining the high precision required in semiconductor fabrication.

3. Real-time Baked Goods Assessment: The model can be utilized in large-scale bakeries to automatically classify and assess the quality of their wafer-based products on the production line, ensuring that only top-quality items are packaged and sold.

4. Research and Development: The AI model can be used in research regarding wafer production and manufacturing processes. It can help enhance current methods by providing quick, accurate analysis of different wafer classes.

5. Training and Education: The model can serve as a practical tool for teaching students or new employees about the different types of wafers and quality standards in the manufacturing industry. It could help users visualize real-world examples of various wafer classes.

Micro-FTIR Filter Images for Particle Detection

This dataset consists of annotated images of filters containing particles. The primary objective of this dataset is to serve as training and validation data for developing a particle detection model using computer vision techniques. More specifically, this dataset can be used to train an image segmentation model that can be used with GEPARD (https://pubmed.ncbi.nlm.nih.gov/32436395/) in order to perform efficient particle detection and analysis using Micro-FTIR microscope.

Two kind of samples are used in our case:

• Normal filters, with a low amount of particles and a clear view of the filter
• Saturated filters, where the particles cover almost all the filter

In the first case, particles were annotated easilly as they are clearly visible over the filter. In the second scenario, the most distinguishable particles on the image have been annotated.

Note

In the case of a saturated filters, the correct method would be to collect a spectral image of the entire filter using a FPA detector or similar and then use tools (e.g. sIMPle ) to analyse this image. However, in our scenario such detector was not available, and a semi-random / operator dependant method had to be used in order to select particles or points for scanning.

FTC SAMPLE DISTANCE DETECTOR

## Overview

FTC SAMPLE DISTANCE DETECTOR is a dataset for instance segmentation tasks - it contains FTC SAMPLES annotations for 2,518 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Here are a few use cases for this project:

1. Lab Sample Analysis: Use the 6R4NU1E2 model in laboratory settings to automatically classify and quantify different gran classes within samples. This could help scientists and researchers streamline their analysis process and improve accuracy in results.

2. Quality Control in Manufacturing: Utilize the model to monitor granular materials on manufacturing lines, such as pharmaceuticals, chemicals, and food production. The model could identify inconsistencies or deviations in gran classes to ensure product quality and safety.

3. Environmental Research: Apply the model in environmental studies for identifying and understanding gran composition in various contexts, such as soil samples, sediment structures, and air particulate matter. This could provide valuable data for both understanding environmental changes and forming policy decisions.

4. Art Restoration and Analysis: Leverage the 6R4NU1E2 model to identify gran classes within paint, pigments, or other artistic media. This information could be used to determine the origin or authenticity of a piece, as well as aid in restoration efforts for historical works of art.

5. Archaeological Studies: Employ the model to analyze gran material found in archaeological digs, helping identify and classify various types of human-made and natural objects. The model could contribute valuable insights into ancient cultures, societies, and methods of crafting materials.

This project aims to develop an automated system for detecting and classifying wood particles recovered from recycling centers. The primary objective is to enhance the sorting process of waste wood by replacing manual methods with a real-time, computer vision–based approach powered by deep learning.

Through analysis of samples collected from recycling facilities, five distinct material types were identified and labeled as separate classes for training and evaluation. The system is built using the YOLOv11m-seg model, which enables instance segmentation for precise identification of particle types.

Ultimately, this project contributes to more efficient and scalable wood recycling workflows by leveraging AI for improved material characterization and classification.

Here are a few use cases for this project:

1. Gardening Applications: The "flowers_segmentation" model could be used in gardening applications. Users could take photos of flowers they are interested in to identify the species and learn more about how to plant and care for them.

2. Biodiversity Studies: Researchers studying biodiversity in particular areas could use this model to help identify and catalogue flower species in various ecosystems.

3. Educational Purposes: The model could be used in educational settings, for example in a botany or biology class, where students need to identify and learn about different types of flowers.

4. Commercial Nursery: Nurseries can use this model to catalog their stock and help customers select plants. It could also help employees identify flowers that come without labels or have lost them.

5. Art Inspiration: Artists might use this model to identify flowers in their environment to learn their names and details, which could provide inspiration for creative works. This might be particularly useful for botanical illustrators or other artists who often work with floral themes.

Here are a few use cases for this project:

1. Fertility Analysis: Medical professionals and researchers could apply the 'Bacchus' model for evaluating male fertility conditions by analysing sperm health in semen samples. The tool could differentiate between healthy sperm cells, unhealthy sperm cells, and debris for an accurate and efficient patient assessment.

2. Educational Tool: This computer vision model can be valuable for biology classrooms when studying human reproduction or cellular structures. By identifying and classifying different spermatoza characteristics, 'Bacchus' can help educators animate their lessons and improve student engagement and comprehension.

3. Research Aid: In a broader sense, researchers in reproductive biology or related fields could utilize 'Bacchus' to automate the classification and analysis of spermatoza, helping to save time and increase accuracy. This aid can contribute to developing new research insights and innovation.

4. Assisted Reproduction Analysis: Fertility clinics could use 'Bacchus' as part of their IVF (In Vitro Fertilization) process. It would aid in selecting the healthiest sperm in sperm donation processes or during sperm collection for IVF methods.

5. Veterinary Reproductive Health: 'Bacchus' could be a game-changer in reproductive health examinations for animals too. The model could be used to determine the fertility condition of breeding animals in the agriculture and pet industry.