Our visitor for final week’s version of OpenCV Weekly Webinar was Gerard Espona of Group Kauda (featured in our first publish). You will discover that episode on YouTube, together with the remainder of the episodes previous and future. This week we had a tutorial on the best way to create your individual panoramas with OpenCV. Signal as much as attend the following webinars at 9am each Thursday on Zoom, free: https://us02web.zoom.us/j/83231402214
On this publish we’re that includes a brief query and reply session with a few of the groups posting cool stuff on-line as a part of OpenCV AI Competitors 2021 utilizing the #OAK2021 hashtag. Due to Groups BenchBotics and UCB_Squirrels for speaking with us this time! Should you missed the earlier editions of this sequence that includes groups and funky movies from the Competitors, return and skim Half 1 and Half 2!
Some Highlights From The #OAK2021 Hashtag
Need your workforce or video to be featured right here? Publish content material with the #OAK2021 hashtag! We’ll undoubtedly see it and so will 1000’s of competitors watchers all around the world.
Group BenchBotics
What’s your undertaking?
BenchBot is a plant phenotyping platform, which consists of two main parts: a semi-automatic picture acquisition gadget and a central course of unit to regulate the platform and digital camera motion. BenchBot is low-cost, modular, upgradeable, moveable, and simple to make use of. Particularly, BenchBot might be adjusted to work in benches of various dimensions and the digital camera peak might be modified to accommodate crops of various heights and sizes. We’re at the moment inspecting picture acquisition and processing approaches that work finest underneath greenhouse situations (i.e. excessive ranges of sunshine and warmth) (Determine 3). Moreover, we’re creating image-processing algorithms wanted to detect and establish crops, detect leaves, decide leaf space and estimate total plant biomass.
Our first prototype consisted of a body that may be moved alongside greenhouse benches with a mounted digital camera system (digital camera and lights) that stops at every potted plant to take particular person RGB and depth photographs. We’re at the moment creating a semi-autonomous system, which requires minimal human intervention. Anticipated merchandise embrace a secure management software and an appropriate database that allows biomass measurements and plant detection algorithms. We’re additionally engaged on full automation of the platform and picture acquisition. We anticipate broad adoption of this know-how amongst private and non-private plant breeders.
Does your workforce have a humorous “origin story?” How did you get collectively?
It’s too sizzling!
Whereas our workforce has been targeted on excessive throughput phenotyping programs for years, this particular method was motivated due to the unbearably sizzling situations that happen in greenhouses within the peak of the summer season. We felt there needed to be a greater approach! In 2019, we explored breeding cowl crops for shade tolerance to suit inside corn and soybean manufacturing programs. This effort started with a greenhouses screening of germplast on the USDA Agricultural Analysis Service web site in Beltsville, Maryland. We had 73 genotypes, 3 remedies, and 4 reps (~876 crops!). As a way to consider the efficiency of the genotypes underneath completely different shade situations, we used traditional phenotyping methods similar to measuring the width and peak of every plant. We additionally took photos of the crops utilizing a PVC body and an ordinary compact digital camera with the intent of utilizing these to guage the efficiency of the crops as properly. This course of was repeated each week for 10 weeks. Measuring and taking photos for nearly 9000 crops, we wanted 2 individuals, greater than 120 hours within the greenhouse in the course of the summer season with greater than 100 F, super-hot. Except you just like the sauna, this isn’t an excellent place to remain for a very long time in the course of the summer season.
How did you determine what drawback to resolve?
The necessity for an automating picture acquisition that would scale back operators’ time within the greenhouse in addition to standardize knowledge acquisition programs utilizing excessive throughput phenotyping birthed BenchBot. in Utilizing easy RGB and depth photographs mixed with a semiautonomous platform at low price may revolutionize the breeding of latest germplasm. Over time, we’ve created enhancements to BenchBot and expanded its purposes to incorporate numerous digital camera programs that correlate to a various array of phenotyping targets in crops.
What’s the most fun a part of #OAK2021 to you?
One of many largest challenges that we face within the growth of this platform is the administration of knowledge move and evaluation in real-time. The OAK-D digital camera might be a game-changer to our undertaking, since we may compress data that’s transmitted to the cloud for additional analyses, thus permitting for environment friendly real-time knowledge move and evaluation. We’re very excited to create a low-cost, semi-automated, non-invasive methodology for figuring out plant species and quantifying their biomass, density, and leaf space underneath greenhouse situations utilizing the brand new OpenCV AI Package (OAK-D) and the Intel Distribution of OpenVINO Toolkit. The profitable execution of this undertaking will tremendously improve entry to high-throughput phenotyping options for breeders, thus accelerating crop enhancements globally.
What do you suppose/really feel upon studying you had been chosen for Section 2?
We’re thrilled to be thought-about for the following section of this competitors. OpenCV is pioneering the best way pc imaginative and prescient and studying is utilized within the know-how sector and such a contest stands to make a constructive influence in quite a few sectors and tendencies in picture processing know-how and purposes of pc imaginative and prescient. It’s an honor and thrilling to take part on this competitors and assist innovate across the purposes of pc imaginative and prescient and machine studying.
What, if something, has shocked you to this point concerning the competitors?
What has been so inspiring for us is the best way our workforce has come collectively and expanded synergies into different workflows. The competitors actually lit a hearth underneath our partnerships and had tremendously accelerated our progress and improvements across the growth of BenchBot.
We considerably anticipated this dynamic as we had the identical expertise within the final competitors. The response time of the help workforce may be very quick and intensive.. It’s nice to enter a contest the place you’ll have the help it’s essential maintain going. We’re very grateful and impressed with the help and efforts by Luxonis.
Do you’ve gotten any phrases on your fellow rivals?
Transferring ahead as a neighborhood is paramount. It’s one of the best ways to develop the know-how, by supporting and inspiring one another. We’re a part of a privileged group of builders, innovators, and customers of know-how and collectively we are able to broaden the worth and use of real-world options to all.
The place ought to readers observe you, to finest sustain together with your progress?
One of the simplest ways to observe our progress is:
Group UCB_Squirrels
What’s your undertaking? Briefly describe your drawback assertion and proposed resolution
Conventional identification programs are restricted because of the progress of safety issues, great amount of data to be monitored, and the latest demand for carrying face masks – which makes face recognition unsuitable. To beat this challenge, we suggest to establish individuals by solely utilizing their approach of strolling since gait traits might be acquired with out important difficulties and from giant distances – all contemplating real-time responses. This know-how is taken into account as essentially the most acceptable biometric for surveillance programs and can be utilized to establish individuals for purposes like entry management and reidentification, amongst others.
Does your workforce have a humorous “origin story?” How did you get collectively?
Most of us have been engaged on Laptop Imaginative and prescient and Machine Studying for some time at UCB. However the place to begin for our workforce on this Competitors might be traced to 1 Fb publish that included the important thing phrases: deadline prolonged. We had few days to construct our proposal and perceive the OAK-D, tough however not inconceivable. Thankfully, we had been engaged on the sector for some years, so it was not an issue to collect our concepts and write a proposal on time. The other occurred with our workforce’s title. Regardless of having a time working collectively, pondering of a reputation was a extremely new activity that nobody anticipated.
How did you determine what drawback to resolve?
Because of the COVID-19 pandemic throughout the globe, Governments have taken measures such because the obligatory carrying of face masks public locations and social distancing. These new insurance policies posed a problem for conventional identification programs that used the face or fingerprint as biometric trait. Consequently, identification programs based mostly on new biometrics, similar to gait, grew to become much more enticing. Moreover, our foremost motivation depends on human identification to keep away from unlawful human trafficking, which is a latent drawback in creating nations like ours.
What’s the most fun a part of #OAK2021 to you?
We might say that essentially the most thrilling factor about this competitors is with the ability to be a part of one thing that’s occurring proper now and should have profound implications on how Laptop Imaginative and prescient is being carried out in real-world situations. Moreover, taking a part of a specific group that may have a hands-on expertise with a state-of-the-art gadget is inspiring us to maintain researching and creating.
What do you suppose / really feel upon studying you had been chosen for Section 2?
At first, we received excited to know that the issue that we had been making an attempt to resolve was thought-about of excessive significance by individuals with years of expertise within the CV area. However then, we additionally got here to understand that being chosen for this section meant that we had been going to have the ability to work OAK-D units and earn first-hand expertise on state-of-the-art units. This blew our minds and received us obsessed with what was coming forward.
What, if something, has shocked you to this point concerning the competitors?
Two issues have shocked us: the pliability/capabilities of the OAK-D, and the help that’s continuously offered. There’s all kinds of tasks that may be accomplished with this gadget, and watching the variety of concepts that the rivals have is enriching and insightful for us. Moreover, we had been additionally shocked with the OAK-D capabilities, as a result of with the ability to course of synthetic intelligence duties contained in the gadget is handy and a really desired property. Lastly, we weren’t anticipating the quantity of help via the OpenCV devoted course and the discord server.
Do you’ve gotten any phrases on your fellow rivals?
Sustain with the nice work, we’ve seen a few of your tasks and are superior. We’re amazed of the huge variety of tasks that might be made through the use of the identical gadget.
The place ought to readers observe you, to finest sustain together with your progress?
Extra To Come
Thanks for studying this third publish in our sequence of workforce profiles. These are just some of the over 200 groups collaborating on this enormous competition- we want all of them the easiest of luck! Should you’re an AI creator who needs to hitch in on the cool stuff, why not purchase your self an OAK-D from The OpenCV Retailer?
Keep tuned for extra profiles, and observe the #OAK2021 tag on Twitter, LinkedIn, or Instagram for a gentle stream of superior stuff from these superb groups. Don’t neglect to join the OpenCV Publication to be notified when new posts go reside, and get unique reductions and gives from our companions.
