This initiative is targeted to support AI/ML developers and enthusiasts for prototyping next-gen autonomous solutions.
Shoot an email at firstname.lastname@example.org if you have specific requirements.
Detection of cattle weight is essential for determining the volume of livestock inputs such as feed or medicine, understanding grade of cattle, and assessing price. Detecting weight is particularly problematic in the grassroots where weighing scales are not available along with having numerous practical challenges.
The beauty of the model we made is that we geared it to be used through low-cost smartphone cameras leveraging reference objects. The model is trained through a dataset of 12,000 cattle imagery and extracted a result of 92% accuracy in detecting weight in controlled environments. And yes, this is completely open source!
Check out our open datasets, ripe for prototyping autonomous systems for AI/ML use-cases in transportation, agriculture, healthcare, sports, spatial intelligence, environment, manufacturing, among others.
320 frames taken Kentucky Derby Highlights from 2019 to 2022 with horses being tracked per frame – identifying race succession. The dataset is appropriate for training computer vision- and data science-based predictions in horse races to complement betting systems.
Labelling Type: Bounding Box, Tracking by FrameDownload.
The 100 frames are taken at every 12th frame (with some blurred frames and outliers replaced) from the match between Real Madrid and Manchester United from open media. The dataset is appropriate for training detection models in respect to sports analytics, of course biased towards soccer.
Labelling Type: Semantic Segmentation, 11 ClassesDownload.
100 images taken from Google Earth Pro appropriate for training spatial and computer vision-based detection models focused on urban mobility and traffic concentrations. The source data was collected from from satellite imagery available in Google Earth Pro. We collected this particular dataset from Edogawa, Tokyo in Japan.
Labelling Type: Bounding Box, 10 Vehicle ClassesDownload.
The motorcycle ride dataset is a collection of 200 frames taken from open media available on YouTube to enable testing for object detection and/or mobility-centered AI solutions - specifically on computer vision powered motorcycle helmets or other inventive avenues.
Labelling Type: Semantic Segmentation, 6 ClassesDownload.
298 frames taken at every 12th frame (with some blurred frames and outliers replaced) from the match between India and Zimbabwe from open media. The dataset is appropriate for training detection models in respect to sports analytics, of course biased towards cricket.
Labelling Type: Semantic Segmentation, 9 ClassesDownload.
Get high-quality and pixel-accurate labelled data through us. We bring the best-in-class, scalable, and adaptive annotation and quality assurance services. Begin the final trek towards fine-tuning your supervised learning model with us.GET IN TOUCH