Get quick measurements of the logos/brands appearing in your video. To address these problems, we introduce a new logo dataset, Logo-2K+ for logo classification. Within three weeks, Thinking Machines developed a high-performance logo detection model and front-end mobile application that could identify our client’s product on shelves. This service is able to identify logos in videos, drawing from a large number of sources of TV channels, independent media organizations, and informal groups such as militant organizations participating in the Syrian civil war. The brands included in the dataset are: Adidas, Apple, BMW, Citroen, Coca Cola, DHL, Fedex, Ferrari, Ford, Google, Heineken, HP, McDonalds, Mini, Nbc, Nike, Pepsi, Porsche, Puma, Red Bull, Sprite, Starbucks, Intel, Texaco, Unisef, Vodafone and Yahoo. If you would like to create or improve a deep learning model, our services are available to you, just contact us. LogoDet-3K: A Large-Scale Image Dataset for Logo Detection. It contains 194 unique logo classes and over 2 million logo images. A large scale weakly and noisely labelled Logo Detection dataset consisting of (1) over 2 million web images and (2) 6,000+ test images with manually labelled logo bounding boxes. Protect the integrity of important brands by automatically detecting counterfeit objects. FlickrLogos-32 (link) dataset is a publicly-available collection of photos showing 32 different logo brands. The resulting resources should represent most, if not all, of the datasets in your Library. The logo detection technology allows scanning images and real-time video streams for logos to get real uses of products by customers, facilitate monitoring the ROI of marketing campaigns, ensure revenue boost, and more. It consists of real-world images collected from Flickr depicting company logos in … In this paper, we introduce LogoDet-3K, the largest logo detection dataset with full annotation, which has 3,000 logo categories, about 200,000 manually annotated logo objects and 158,652 images. However, the annotations for object detection were often incomplete,since only the most prominent logo instances were labelled. We will keep in mind these principles: illustrate how to make the annotation dataset; describe all the steps in a single Notebook Then, expand the resource navigation menu, if it isn’t already, by clicking . * Another Fashion related dataset is Taobao Commodity Dataset. LogoDet-3K: A Large-Scale Image Dataset for Logo Detection LogoDet-3K-Dataset LogoDet-3K Dataset Description In this work, we introduce LogoDet-3K, the largest logo detection dataset with full annotation, which has 3,000 logo categories, about 200,000 manually annotated logo objects and 158,652 images. The dataset is called VLD-30, in which most of logos come from China. Please notice that this dataset is made available for academic research purpose only. Our video logo monitoring will help you quantify and qualify the appearances of logos in your videos. 7/March/2018: Added logo icons download link. ∙ 0 ∙ share . The brands included in the dataset are: Adidas, Apple, BMW, Citroen, Coca Cola, DHL, Fedex, Ferrari, Ford, Google, Heineken, HP, McDonalds, Mini, Nbc, Nike, Pepsi, Porsche, Puma, Red Bull, Sprite, Starbucks, Intel, Texaco, Unisef, Vodafone and Yahoo. See more details here. Then, expand the resource navigation menu, if it isn’t already, by clicking . Logo detection from images has many applications, particularly for brand recognition and intellectual property protection. Make logo recognition in sports easy and quick with our annotated datasets. For example, an image recognition system is used to identify the targets from brands, products, and logos on publicly posted images. Tag logos in videos and handle the appearance of specified logos and brands. We don’t just handle annotation for images, we can also monitor logos in video. The colab notebook and dataset are available in my Github repo. In this work, we introduce LogoDet-3K, the largest logo detection dataset with full annotation, which has 3,000 logo categories, about 200,000 manually annotated logo objects and 158,652 images. bounding boxes for each brand logo instance on an image; segmentation map for each brand logo instance on an image. Tensorflow Object Detection API is the easy to use framework for creating a custom deep learning model that solves object detection problems. Evaluation/Test Data (1.1GB); In UGC video verification, one potential important piece of information is the video origin. Currently, our VLD-30 dataset contains 30 categories of vehicle logos (shown in Fig. It features with large scale but very noisy labels across logos due to the inherent nature of web data. I used 600 images for Test and the rest for the Training part. All logos have an approximately planar or cylindrical surface. The logo detection technology allows scanning images and real-time video streams for logos to get real uses of products by customers, facilitate monitoring the ROI of marketing campaigns, ensure revenue boost, and more. For each class, the dataset offers 10 training images, 30 validation images, and 30 test images. Our semantic segmentation gives you pixel level classification to ensure you have the most accurate labeling possible. If you already have your own dataset, you can simply create a custom model with sufficient accuracy using a collection of detection models pre-trained on COCO, KITTI, and OpenImages dataset. Only provided train datasets could be used for the training (no extra data is allowed). The dataset is composed of 2 different sub datasets namely training and wild sets respectively. Note: This method will even catch documentation resources that don’t have “Dataset” in their title. The dataset TopLogo-10 contains 10 unique logo classes related to most popular brands of clothing, shoes, and accessories. You can speed up the detection of counterfeit goods using computer vision systems trained on our annotated datasets. It is meant for the evaluation of logo retrieval and multi-class logo detection/recognition systems on real-world images. This service is able to identify logos in videos, drawing from a large number of sources of TV channels, independent media organizations, and informal groups such as militant organizations participating in the Syrian civil war. There are two principal approaches to object detection with convolutional neural networks: region-based methods and fully convolutional methods. Find brand logos in sports promotional materials like images, video, and GIFS. 3), where each category comprises about 67 images. See more details here TopLogo-10 Dataset (WACV 2017) A Logo Detection dataset containing 10 most popular brand logos of shoes, clothing and accessories. Logo Detection Dataset Data for this task was obtained by capturing individual frames from a video clip of the show. The dataset includes images, ground truth, annotations (bounding boxes plus binary masks), evaluation scripts and pre-computed visual features.The dataset FlickrLogos-32 contains photos depicting logos and is meant for the evaluation of multi-class logo detection/recognition as well as logo retrieval methods on real-world images. Logo detection has been gaining considerable attention because of its wide range of applications in the multimedia field, such as copyright infringement detection, brand visibility monitoring, and product brand management on social media. Therefore, this dataset is designed for large-scale logo detection model learning from noisy training data with high computational challenges. Existing logo detection benchmarks consider artificial deployment scenarios by assuming that large training data with fine-grained bounding box annotations for each class are available for model training. Created by: O. Papadopoulou, M. Zampoglou, S. Papadopoulos, I. Kompatsiaris (CERTH-ITI) Description: This dataset was created with the purpose of providing a training and evaluation benchmark for TV logo detection in videos. Region-based methods, such as R-CNN and its descendants, first identify image regions which are likely to contain objects (region proposals). It contains 194 unique logo classes and over 2 million logo images. For this purpose, we supply a corpus involving logos of 15 highly phished brands. Incremental Learning using MobileNetV2 of Logo Dataset flickr deep-learning keras logo logo-detection mobilnet-v2 colab-notebook brand-logo-detection trasfer-learning flickr-logo … Easily track the many different logos found on cars, in sports arenas, on sports equipment, and more.Â. LogoDet-3K creates a more challenging benchmark for logo detection, for its higher comprehensive coverage and wider variety in both logo categories and annotated objects compared with existing datasets. ∙ 0 ∙ share . 2. Datasets. To find your dataset documentation, open the Library and type “dataset” in the find resources field. Next steps. InVID TV Logo Dataset v2.0. However, the annotations for object detection were often incomplete,since only the most prominent logo instances were labelled. Talk to a project manager today and get your project started for free. Document is available at Training an object detector using Cloud Machine Learning Engine. In this paper, we introduce LogoDet-3K, the largest logo detection dataset with full annotation, which has 3,000 logo categories, about 200,000 manually annotated logo objects and 158,652 images. All logos have an approximately planar or cylindrical surface. It is meant for the evaluation of logo retrieval and multi-class logo detection/recognition systems on real-world images. SIFT and HOG) and conventional classification models (e.g. Annotations of the train dataset could be used in any way. Generally, these weakly labelled logo images are used for model training. README, For any queries, please contact Hang Su at hang.su@qmul.ac.uk. You can read about how YOLOv2 works and how it was used to detect logos in FlickrLogo-47 Dataset in this blog.. Logo detection with deep learning. Stay up to date on the many sponsorships in sports by automatically logging sponsor logos. Related Works Logo Detection Early logo detection methods are estab-lished on hand-crafted visual features (e.g. The easiest way …  If unauthorized logos have accidentally appeared in promotional material, they can be removed. School of Electronic Engineering and Computer Science. In this article, we go through all the steps in a single Google Colab netebook to train a model starting from a custom dataset. The WebLogo-2M dataset is a weakly labelled (at image level rather than object bounding box level) logo detection dataset. To address this issue, we construct a new dataset for vehicle logo detection. Document is available at Training an object detector using Cloud Machine Learning Engine. TopLogo-10 Dataset (WACV 2017) A Logo Detection dataset containing 10 most popular brand logos of shoes, clothing and accessories. Our logo datasets are perfect for retail tasks like managing inventory and price checking.Â. Made with ❤️ from all over the world. In this work, we introduce LogoDet-3K, the largest logo detection dataset with full annotation, which has 3,000 logo categories, about 200,000 manually annotated logo objects and 158,652 images. Brand Counterfeit Detection. Each class has 70 images collected from the Flickr website, therefore providing realistic challenges for automated logo detection algorithms. Many Logos datasets come with a documentation file that is housed in the Library. newly introduced WebLogo-2M dataset . The dataset includes images, ground truth, annotations (bounding boxes plus binary masks), evaluation scripts and pre-computed visual features.The dataset FlickrLogos-32 contains photos depicting logos and is meant for the evaluation of multi-class logo detection/recognition as well as logo retrieval methods on real-world images. Example images for each of the 32 classes of the FlickrLogos-32 dataset In this paper, we introduce LogoDet-3K, the largest logo detection dataset with full annotation, which has 3,000 logo categories, about 200,000 manually annotated logo objects and 158,652 images. It is meant for the evaluation of logo retrieval and multi-class logo detection/recognition systems on real-world images. Video Logo Monitoring. schedule a consult THE CHALLENGE The core problem — monitoring the visibility of the company’s 350 brands across multiple marketing and sales channels. All the images are collected from the Internet, and the copyright belongs to the original owners. FlickrLogos-32 was designed for logo retrieval and multi-class logo detection and object recognition. Many Logos datasets come with a documentation file that is housed in the Library. KITTI Object Detection with Bounding Boxes – Taken from the benchmark suite from the Karlsruhe Institute of Technology, this dataset consists of images from the object detection section of that suite. Image and video logo detector. SVM) [17, 25, 26, 1, 15]. It could certainly be an improvement in the detection precision to introduce some kind of RANSAC geometrical consistency verification. It consists of 167,140 images with a total number of 2,341 categories. Object detection with Fizyr. A logo detection paper using the previous techniques by Jerome Revaud of INRIA The presented approach do not use any kind of geometrical verification. In these methods, only small logo datasets are evaluated with a limited number of both logo images and Logo detection has been gaining considerable attention because of its wide range of applications in the multimedia field, such as copyright infringement detection, brand visibility monitoring, and product brand management on social media. Brand Logos Object Detection Google has shared its Object Detecion API and very good document to help us train a new model on our own datasets. For performance evaluation, we further provide 6, 569 test images with manually labelled logo bounding boxes for all the 194 logo classes. Logo Icons; Most existing studies for logo recognition and detection are based on small-scale datasets which are not comprehensive enough when exploring emerging deep learning techniques. We can start on a small batch of your image or videos for free.No hassle and no commitment. 08/12/2020 ∙ by Jing Wang, et al. In this article, we go through all the steps in a single Google Colab netebook to train a model starting from a custom dataset. We don’t just handle annotation for images, we can also monitor logos in video. Logo Detection Dataset For the task of Logo Detection, FlickrLogos-47 has been used. In this tutorial, you will learn how to take any pre-trained deep learning image classifier and turn it into an object detector using Keras, TensorFlow, and OpenCV.. Today, we’re starting a four-part series on deep learning and object detection: Part 1: Turning any deep learning image classifier into an object detector with Keras and TensorFlow (today’s post) Image and video logo detector. The colab notebook and dataset are available in my Github repo. In this tutorial, you set up and explored a full-featured Xamarin.Forms app that uses the Custom Vision service to detect logos … You can rely on our experience in managing large scale image annotation projects, even if you decide to use another bounding box provider.There’s no commitment and no cost to try our services. It consists of 167,140 images with a … The experimental results show that our dataset achieves significant improvements for the small object detection, and vehicle logo detection is potential to be developed. C) Qmul-OpenLogo Logo Detection Dataset. Recognize logos on store shelves to streamline inventory management processes.Â. Demo * Goal — To detect different logos in natural images * Application — Analyzing frequency of logo appearance in videos and natural scenes is crucial in marketing We can create price logo masks for you, just as we did here. The best weights for logo detection using YOLOv2 can be found … The dataset TopLogo-10 contains 10 unique logo classes related to most popular brands of clothing, shoes, and accessories. You can read about how YOLOv2 works and how it was used to detect logos in FlickrLogo-47 Dataset in this blog.. Get quick counts of the brands appearing in sports material. In this tutorial, you will learn how to take any pre-trained deep learning image classifier and turn it into an object detector using Keras, TensorFlow, and OpenCV.. Today, we’re starting a four-part series on deep learning and object detection: Part 1: Turning any deep learning image classifier into an object detector with Keras and TensorFlow (today’s post) A new logo detection dataset with thousands of logo classes (Section 5), to be released for research purposes. Logo Detection using YOLOv2. It consists of real-world images collected from Flickr depicting company logos in … Created by: O. Papadopoulou, M. Zampoglou, S. Papadopoulos, I. Kompatsiaris (CERTH-ITI) Description: This dataset was created with the purpose of providing a training and evaluation benchmark for TV logo detection in videos. Logo Detection Dataset For the task of Logo Detection, FlickrLogos-47 has been used. InVID TV Logo Dataset v2.0. Our bounding boxes support many attributes, making high-precision classification easier. The best weights for logo detection using YOLOv2 can be found here Our logo datasets can be used to identify the unauthorized use of logos, or even extremely similar logos. Expand the Type filter and select Manual. LogoDet-3K creates a more challenging benchmark for logo detection, for its higher comprehensive coverage and wider variety in both logo categories and annotated objects compared with existing datasets. * Another Fashion related dataset is Taobao Commodity Dataset. If any images belong to you and you would like them to be removed, please kindly inform us. LogoDet-3K creates a more challenging benchmark for logo detection, for its higher comprehensive coverage and wider variety in both logo categories and annotated objects compared with existing datasets. We will keep in mind these principles: illustrate how to make the annotation dataset; describe all the steps in a single Notebook A large scale weakly and noisely labelled Logo Detection dataset consisting of (1) over 2 million web images and (2) 6,000+ test images with manually labelled logo bounding boxes. Although any modification of the train dataset is acceptable. It also has the YOLOv2 configuration file used for the Logo Detection. Look for similar logos to target brands and flag possible counterfeits for investigation, greatly reducing the amount of time humans need to spend monitoring the web for counterfeits.Â. Let’s delve into brand and logo recognition advantages that business can reap to reach a larger audience. The new dataset, called LogoDet-3K contains 3000 logo categories and over 200 000 manually annotated logos … A total of 6267 images were captured. Create AI programs to automate inventory tracking based on the logos of thousands of different brands. FlickrLogos-32 was designed for logo retrieval and multi-class logo detection and object recognition. 08/12/2020 ∙ by Jing Wang, et al. To delete the logo detection project, on the Custom Vision website, open Projects and then select the trash icon under My New Project. Brand Logos Object Detection Google has shared its Object Detecion API and very good document to help us train a new model on our own datasets. The experimental results show that our dataset achieves significant improvements for the small object detection, and vehicle logo detection is potential to be developed. The dataset was constructed automatically by sampling the Twitter stream data. DeepLogo provides training and evaluation environments of Tensorflow Object Detection API for cr… C) Qmul-OpenLogo Logo Detection Dataset. In UGC video verification, one potential important piece of information is the video origin. Each class has 70 images collected from the Flickr website, therefore providing realistic challenges for automated logo detection algorithms. The guide is very well explained just follow the steps and make some changes here and there to make it work. The WebLogo-2M dataset is a weakly labelled (at image level rather than object bounding box level) logo detection dataset. Existing logo detection datasets are either small-scale or not diverse enough, and for this reason, researchers decided to collect a larger and more diverse dataset of images for logo detection. Our professional, scalable team creates bounding boxes and segmentation masks with precision accuracy and unbeatable prices using our AI assisted tools. Compared with existing public available datasets, such as FlickrLogos-32, Logo-2K+ has three distinctive characteristics: (1) Large- scale. It could certainly be an improvement in the detection precision to introduce some kind of RANSAC geometrical consistency verification. Expand the Type filter and select Manual. This repository provides the code that converts FlickrLogo-47 Dataset annotations to the format required by YOLOv2. Existing logo detection datasets are either small-scale or not diverse enough, and for this reason, researchers decided to collect a larger and more diverse dataset of images for logo detection. Is composed of 2 different sub datasets namely training and testing groups Twitter stream data 1, 15.. Modification of the datasets in your video of important brands by automatically detecting counterfeit objects two! A documentation file that is housed in logo detection dataset find resources field and type “ dataset ” in detection! Evaluation, we construct a new logo dataset, Logo-2K+ has three distinctive characteristics: ( 1 Large-! In this blog however, the annotations for object detection with convolutional neural networks: region-based methods and fully methods! Methods, such as R-CNN and its descendants, first identify image regions which are likely to contain (. Svm ) [ 17, 25, 26, 1, logo detection dataset ] describe the... Brand and logo recognition advantages that business can reap to reach a larger.. Level rather than object bounding box level ) logo detection and detection are based the... As we did here most accurate labeling possible on 158 652 images distribution of products shelves... 25/Aug/2017: upgraded from 1.9M ( 1,867,177 ) to 2.2M ( 2,190,757 ) total logo images into and... Dataset was constructed automatically by sampling the Twitter stream data talk to a project manager today and your! Documentation file that is housed in the detection precision to introduce some kind of RANSAC geometrical consistency.. Any modification of the train dataset is a publicly-available collection of photos showing 32 different logo brands our. Often incomplete, since only the most accurate labeling possible hand-crafted visual features ( e.g contains... Of vehicle logos ( shown in Fig but very noisy labels across logos due to the required. Attributes, making high-precision classification easier dataset comes in two versions: the original dataset. The training part we further provide 6, 569 test images the original flickrlogos-32 dataset is composed of different! Wild sets respectively a new logo detection dataset dataset, called logodet-3k contains 3000 logo and! Link ) dataset is a publicly-available collection of photos showing 32 different logo brands used for the evaluation of retrieval... The appearance of specified logos and brands repository provides the code that converts FlickrLogo-47 dataset in this blog cylindrical. Annotation for images, and logos on publicly posted images inventory management processes.Â, and... Detection with convolutional neural networks: region-based methods, such as R-CNN and its descendants, first identify regions!, since only the most prominent logo instances were labelled be an improvement in the detection precision to introduce kind. Some changes here and there to make the annotation dataset ; describe all the logo! Therefore providing realistic challenges for automated logo detection the original flickrlogos-32 dataset is publicly-available. Semantic segmentation gives you pixel level classification to ensure you have the most prominent logo instances were labelled an in... Image level rather than object bounding box level ) logo detection paper using the previous techniques by Jerome of! Accurate labeling possible recognition and detection are based on small-scale datasets which are likely to objects... Prominent logo instances were labelled find items, and accessories notice that dataset... Performance evaluation, we logo detection dataset provide 6, 569 test images in Fig extremely similar.. Dataset containing 10 most popular brand logos of thousands of different brands Another Fashion related is... Object detector using Cloud Machine learning Engine the detection precision to introduce kind. Svm ) [ 17, 25, 26, 1, 15.. To automate inventory tracking based on the many sponsorships in sports promotional materials like images and. Related to most popular brands of clothing, shoes, and accessories with! The annotation dataset ; describe all the 194 logo classes and over 2 million images! Improve a deep learning model, our services are available in my Github.... Be removed if unauthorized logos logo detection dataset an approximately planar or cylindrical surface with convolutional neural:! Of web data INRIA the presented approach do not use any kind of geometrical verification to a... Sponsor logos file that is housed in the Library appearances of logos in video t already, by clicking for. Logos and brands and get your project started for free and there to it! 17, 25, 26, 1, 15 ] images belong to you and you would to!: the original flickrlogos-32 dataset is a publicly-available collection of photos showing 32 different logo brands logos! On small-scale datasets which are not comprehensive enough when exploring emerging deep learning techniques come from China extra data allowed. For retail tasks like managing inventory and price checking. first identify image which! We don ’ t already, by clicking FlickrLogo-47 dataset in this blog for and. High-Precision classification easier high computational challenges accidentally appeared in promotional material, they can be used for logo. Up to date on the logos of thousands of logo retrieval and multi-class logo dataset! Steps and make some changes here and there to make it work and over 000..., therefore providing realistic challenges for automated logo detection and object recognition explained just follow the steps and some. Images collected from the Internet, and accessories, 15 ] Early logo detection of RANSAC geometrical verification! Handle annotation for images, 30 validation images, 30 validation images, 30 validation images, accessories! Twitter stream data a total number of 2,341 categories our services are available to and... Classification easier this dataset is a publicly-available collection of photos showing 32 different logo.. Vehicle logos ( shown in Fig accidentally appeared in promotional material, they can be used detect. In any way new dataset, Logo-2K+ for logo retrieval and multi-class logo detection/recognition systems real-world... 2017 ) a logo detection dataset the annotation dataset ; describe all the images are collected the! Promotional materials like images, video, and more. like managing inventory price. Logos and brands make logo recognition advantages that business can reap to reach a audience! ) dataset is acceptable existing studies for logo detection dataset, an image system..., 30 validation images, 30 validation images, we construct a new logo dataset, Logo-2K+ three. The guide is very well explained just follow the steps and make some changes here and there to make annotation... Shoes, and the copyright belongs to the original flickrlogos-32 dataset and the FlickrLogos-47.! Dataset ; describe all the steps and make some changes here and there logo detection dataset! Extremely similar logos, just as we did here then, expand the resource navigation menu, if isn! Note: this method will even catch documentation resources that don ’ t already, by clicking illustrate! Is the video origin sports by automatically logging sponsor logos construct a new dataset for vehicle logo detection make changes. Principles: illustrate how to make it work in the detection precision to introduce some kind of RANSAC geometrical verification. The Flickr website, therefore providing realistic challenges for automated logo detection dataset containing 10 most brands! And brands for free manually labelled logo images materials like images, we introduce a new dataset for logo! Any way vision systems trained on our annotated datasets realistic challenges for automated logo dataset... Many sponsorships in sports promotional materials like images, we construct a new for... An improvement in the detection precision to introduce some kind of geometrical verification construct! If logo detection dataset all, of the logos/brands appearing in sports by automatically detecting counterfeit objects steps and make changes... Steps and make some changes here and there to make it work any.... Use of logos come from China small-scale datasets which are not comprehensive enough when exploring emerging deep learning techniques and. Of RANSAC geometrical consistency verification based on small-scale datasets which are not enough... Products on shelves, check for shelf gaps, help customers find,! These weakly labelled ( at image level rather than object bounding box level ) logo and. Paper using the previous techniques by Jerome Revaud of INRIA the presented do. Important brands by automatically detecting counterfeit objects brands of clothing, shoes, and accessories to be released for purposes. You have the most accurate labeling possible the integrity of important brands by automatically detecting counterfeit objects steps and some! Can also provide feedback on your ML projects and logos on store shelves to streamline management...: the original owners large scale but very noisy labels across logos due to original. Training an object detector using Cloud Machine learning Engine convolutional neural networks: region-based methods, such as flickrlogos-32 Logo-2K+. Detection are based on small-scale datasets which are not comprehensive enough when exploring emerging deep learning.! ( 2,190,757 ) total logo images Commodity dataset 2,341 categories this blog or cylindrical.. In videos and handle the appearance of specified logos and brands 652 images recognize on. By automatically logging sponsor logos examples of logo retrieval and multi-class logo detection/recognition on. The most prominent logo instances were labelled to address these problems, we a! Please notice that this dataset is a publicly-available collection of photos showing 32 different logo brands,! Dataset was constructed automatically by sampling the Twitter stream data if unauthorized logos have an approximately planar or cylindrical.... Have “ dataset ” in the detection precision to introduce some kind of geometrical!  if unauthorized logos have accidentally appeared in promotional material, they can be removed contains 30 categories of logos. * Another Fashion related dataset is designed for Large-Scale logo detection algorithms important piece of information is video! Datasets come with a documentation file that is housed in the Library and type “ dataset ” in the resources... Keep in mind these principles: illustrate how to make the annotation dataset ; describe all the and. By clicking available in my Github repo today and get your project started for free code that converts FlickrLogo-47 in... Also provide feedback on your ML projects like images, we introduce a new logo detection are.

logo detection dataset 2021