Synativ provides a platform for training computer vision models without the need to collect and label large amounts of images.

Synativ

We believe engineers need the right tools to accelerate their computer vision development. Leverage visual foundation models to turn small datasets into working models with just a few lines of code. No need to wait for data acquisition, data labelling, or R&D teams. Spending too much time on CV - reach out :)

We are backed by Sequoia Capital (Arc) and are always looking for more co-innovators, partners, and folks eager to join our team!

We are hiring: https://synativ.freshteam.com/jobs

Synativ’s Earth Observation Week 🛰 🌍 – DAY 5

Too many serious posts this week. Here is a cool Google Earth find instead. Apparently made from actual coke bottles in the Mojave desert (at 18.5292 S, 70.2500 W). Great use of earth observation by The Coca-Cola Company marketing team 👏 🥂 

#ai #earthobservation #marketing #computervision

Synativ’s Earth Observation Week 🛰 🌍 – DAY 4

3 sentence paper. A lot of academics are working tirelessly to push the field of research at the intersection of AI and Earth Observability. So, we asked our CTO Dr. Tom Bruls to pick his favorite paper this week and explain it in 3 sentences:

📰 Paper: GeoLLM-QA by Microsoft Research https://lnkd.in/gNrfRhE6

- The future workflows are going to be based on realistic use / jobs to be done performed by an agent connected to geospatial tools such as ArcGIS
- The team constructed a new benchmark which can evaluate agents against each other and track progress. carefully thought about the evaluation metrics which is amazing
- The paper focuses on computer vision tasks based on "golden detectors", but we believe they can be replaced by real models (💪 Synativ can help with that)

#earthobservation #geospatial #computervision #deeplearning #foundationmodels #data #geoai #research

Synativ’s Earth Observation Week 🛰 🌍 – DAY 3

We believe that the emergence of visual foundation models in remote sensing will allow to build earth observation AI capabilities much faster and cheaper. 

Jack-bo1220 published a great one-stop-shop for papers 📰, datasets 🖼, benchmarks 📈, code 👨💻, and pre-trained weights ⚖ for Remote Sensing Foundation Models (RSFMs). Fantastic starting point to tap into the power of foundation models - https://lnkd.in/gjRPz6V2 

In addition to having these models, you want to be able to evaluate them . So, we at Synativ started an open source project - RSFMs evaluation playground -  https://lnkd.in/g2RCMjnk 


#earthobservation #geospatial #computervision #deeplearning #foundationmodels #data #geoai

Synativ’s Earth Observation Week 🛰 🌍 – DAY 2

Today, in San Diego, Esri User Conference is going at full steam. It attracts many industry leaders, enthusiasts, users, and vendors all taking all things earth observation. 

🏆 If you are there, here are the top 5 things we are excited about (of course it is all about getting working computer vision and deep learning models on geospatial data):
- Imagery Summit https://lnkd.in/eHJgM78g
- Using Large Language Models and Foundation Models by Rohit S. and Karthik Dutt https://lnkd.in/emHq4ba2
- The Day 3 Problem: Keeping Geospatial Computer Vision Models Performant by Eddy Chavarria https://lnkd.in/esvrnZuT
- GeoAI Unleashed: How Geospatial is Set Free with Deep Learning by Steven P. Santovasi, GISP, CPM, CPMM https://lnkd.in/eThqQKQB
- GeoAI Best Practices by by Rohit S., Karthik Dutt , and Priyanka Tuteja https://lnkd.in/e3Cn4GQx

Thank you for all the sessions leaders putting in time, effort, and energy into putting these topics together.

Let us know if we missed anything!

#earthobservation #geospatial #computervision #deeplearning #foundationmodels #data #esri #geoai

Synativ’s Earth Observation Week 🛰 🌍 – DAY 1

We have had a lot of exciting geospatial customers lately (startups and big players). So, similar to our agriculture week, we will post every day this week to talk about the cool intersection of computer vision and earth observation applications. 

Satellite and aerial imagery hold an incredible amount of information. This data can be hard to get. Here are few of the data sources we like that you can use as you are staring to build earth observation capabilities:

1) https://lnkd.in/gzHdX3d - spacenet datasets contain ~67,000 square km of very high-resolution imagery, >11M building footprints, and ~20,000 km of road labels to ensure that there is adequate open source data available for geospatial machine learning applications
2) https://neo.gsfc.nasa.gov/ - The NASA Earth Observations dataset is great if you are focusing on environmental use cases
3) https://lnkd.in/eEskViSW - this Kaggle challenge is 7 years old but still is a great dataset to build a prototype with (3-band and 16-band data available)

Go build AI models! 🧠 🤖
#earthobservation #geospatial #computervision #deeplearning #foundationmodels #data

Synativ’s Agriculture Week 👨🌾 – DAY 4

For some of you the week is over – enjoy the pubs! 🍺🍺🍺. As a part of highlight challenges and opportunities in agriculture and computer vision – let’s talk about use case creativity. 

Often companies forget that image data is everywhere and can be used to train unexpected computer vision models. Here is a cool example of a team from Imperial College using computer vision to look at beer carbonation (https://lnkd.in/eWnZUEbQ). Beer doesn’t grow outside but hops and barley do so this does fit into agriculture week!. 

Just a reminder to think about data availability and how AI can enhance even unexpected processes! We love IDEO U as it has great design thinking resources to help with that - https://lnkd.in/ehAf_FA6

#ai #computervision #segmentation #usecases

Synativ’s Agriculture Week 👨🌾 – DAY 3

Many Agriculture AI teams struggle to go from prototype to a working computer vision model in production easily. As you can imagine a field  / forest / pasture looks very differently in the spring as opposed to the fall. Here are some best practices to consider as you are planning to have a robust CV model

1) Always double check your test sets. If the test set does not represent deployment conditions – you may spend a lot of cycles fixing errors and edge cases
2) Check model format, deployment environment requirements, and security
3) Have monitoring / evaluation plan. More than half of our users hear about their models issue from their customers.
4) Have a responsible person for post-deployment maintenance (doesn't have to be a full time effort)

#ai #computervision #agriculture #deployment

Synativ’s Agriculture Week 👨🌾  – DAY 2

📰 3 Sentence paper. A lot of academics are working tirelessly to push the field of research. So, this week, we asked our CTO Tom Bruls to pick his favorite paper and explain it in 3 sentences:

Paper: YOLOv1 to YOLOv10: A comprehensive review of YOLO variants and their application in the agricultural domain https://lnkd.in/ex4F27SR

- YOLO has established itself as a “plug-and-play” solution for many object detection tasks.
- This paper evaluates how YOLO has progressed over the years up to the latest v10 for agriculture applications. YOLOv8 developed by Ultralytics is still close to state of the art and later version have mainly focused on efficiency gains
- Edge deployment and small objects are still areas for improvement and will require additional engineering effort (shameless plug - this is something Synativ can help with 💪 )

We recently connected with Muhammad zahid Hussain (who happens to be one of the authors on this paper) and love the work his team is doing in the domain!

#computervision #agriculture #ai #yolo #research

Synativ’s Agriculture Week. 👨🌾  – DAY 1

We have had a lot of happy agriculture customers lately (startups and big players). So, we figured we will post every day this week to talk about the cool intersection of computer vision and agriculture. 

Today let’s talk about our favorite use cases we have seen in the industry so far (in no particular order):

- ☘ Weed vs crop: ability to detect and segment weeds can save millions of dollars on pesticides, machinery, and crop yields. Shoutout to Sentera for making great progress in the space
- 🐛 Crop disease detection: some incredible stories how a missed disease can wipe off whol fields. Companies can use computer vision to detect disease early
- 🚜 Autonomous harvesting: last week we talked to an engineer who picked grapes in his youth. To avoid back problems like he did – it is way cooler to get machines to do the work
- 🐄 Animal / Cattle welfare. Some cool companies out there are using computer vision to count, detect, and even analyze animal behaviour. Shoutout to fellow UK startup CattleEye.
- 🔢 Object Counting. Now, it may seem boring but only to a human. imagine if you can take one image and count number of Christmas lights in a click of a button.  (source:hackaday)

#computervision #ai #usecases

Synativ is hanging out at Machine Vision Conference today. We have had a lot of success in helping companies build defect detection models - looking forward to meeting many more. If you are here - let Andrew Kanapatski  or Tom Bruls  know!

Thanks UKIVA (UK Industrial Vision Association) for organizing

#ai #machinevision #computervision #foundationmodels

COM:SYNATIV

Synativ

Analysis Not Available