10 ms·
YOLO: Real-Time Object Detection
- abhisuri97 9y agoFor the threshold value, how is 0.25 determined to be the most optimal? is this value just specific to the datasets YOLO was trained on or is it something that is universal. Also, this is really awesome!
- nl 9y agoGenerally you learn a threshold value like that under cross validation.
- iampims 9y agoImpressive speed.
- plaguuuuuu 9y ago(a) very cool project (b) ..interesting CV
- bangonkeyboard 9y agoTED talk: https://www.ted.com/talks/joseph_redmon_how_a_computer_learns_to_recognize_objects_instantly https://www.ted.com/talks/joseph_redmon_how_a_computer_learn...
- ricardobeat 9y agoIf you look closely at the video, the quality of detection is not really great for anything besides `person`. I wonder if weighing objects by the rate they are being found over N frames, and conforming to a somewhat linear trajectory could improve results; feeding matches from previous frames could also help find the same object in subsequent ones, and maybe even work out temporary occlusion (thinking out loud here - probably way more complex than one would expect).
- aeleos 9y agoI'm not 100% sure about this but I think part of the reason why its called you only look once is that it is single frame object detection. I do know that there are other types of networks that can previous frames into account, but the models themselves are much larger (in terms of VRAM requirements for loading into memory) and more computationally intensive. This network is special because it can run with very little power. From what I remember it can run on a 10W board at 6fps which compared to networks only a few years ago is 10x decrease
- Animats 9y agoIt's clearly single-frame; you can watch recognition succeed and fail from frame to frame. The "person" recognizer misses some clear faces, so it's not heavily face-oriented. Since it's single-frame, it's not recognizing articulated motion. Recognition seems to be limited to "person", "motorbike", "tie", "cell phone" (a gun, actually), "umbrella", "truck" (misidentified part of a train) "bench" (a railing) and "horse" (motorbike with duffel seen from rear). "Person", "umbrella", "tie", and "motorbike" seem to work; the others are kind of random. The trouble with running recognizers on Hollywood movies is that they have many conventions of what appears on screen and how big it is on screen. Are they training on such data? Good data sets would be side views from a moving vehicle, like Google StreetView data or just a GoPro pointed sideways while driving around.
- fattire 9y agoStill, "person" can be pretty useful/scary. https://twitter.com/Sentdex/status/899057144884015104 https://twitter.com/Sentdex/status/899057144884015104
- taneq 9y agoIt still seems to pick up a lot more objects than some of the video we've seen from, eg. Tesla Autopilot.
- siliconc0w 9y agoHypothesis: All demos are improved with the addition of dragonforce.
- reificator 9y agoDiscovery: Blaring music of any kind, including my preferred genres, is unpleasant when trying to watch a demo.
- jayd16 9y agoDoesn't hold up in independent tests.
- eggie5 9y agomost of the top submissions to the last ImageNet comp. are tweaks on YOLO and Faster RCNN. But why is this trending now?
- Oras 9y agocould be because more people are getting into DL and image recognition? I am one of them so what might be old to some people is new info for beginners
- hedgehog 9y agoThis is good background but modern architecture are significantly better: Learning Transferable Architectures for Scalable Image Recognition https://arxiv.org/abs/1707.07012 https://arxiv.org/abs/1707.07012
- 0xbear 9y agoThis is not "modern" though, the paper is quite old. He has improved it significantly since then: https://arxiv.org/abs/1612.08242 https://arxiv.org/abs/1612.08242. Best paper honorable mention at CVPR2017.
- hedgehog 9y agoI missed that the site was updated for YOLO9000, it is very impressive. The Google paper is recent though (July), maybe you're thinking of https://arxiv.org/abs/1611.10012 https://arxiv.org/abs/1611.10012? Anyway, what I meant to point out was that there has been a lot of work since on improving the efficiency of the architectures used for feature extraction (Xception, MobileNet, ShuffleNet, NASNet).
- 0xbear 9y agoSure, there's new work all the time. But I think you might be missing the forest for the trees here. The main efficiency gain here is not in how you extract and transform features (although those improvements would also work in YOLO, as far as I can tell). The novel part is in structuring the problem as a single forward pass regression. Further improvement in this year's paper is in a better loss function. Other possible efficiency improvements are orthogonal and complementary to that.
- hedgehog 9y agoI agree that the single shot approach in YOLO was pretty clearly a big step forward. In the two years since though there much of the efficiency work has been in the underlying feature detector architecture, which as you point out should integrate well with the YOLO9000 training improvements. It's very cool to see this kind of capability get within range of phone-sized devices.
- deepnotderp 9y agoThis is cool but outdated, there're other better papers now.
- edshiro 9y agoWhat better papers would you recommend reading?
- deepnotderp 9y agoArxiv is thou friend :) https://www.google.com/search?q=fast+object+detection+deep+learning+arxiv&oq=fast+object+detection+deep+learning+arxiv&gs_l=psy-ab.3...7628.7966.0.8024.5.4.0.0.0.0.102.315.3j1.4.0.foo%2Cnso-ehuqi%3D1%2Cnso-ehuui%3D1%2Cewh%3D0%2Cnso-mplt%3D2%2Cnso-enksa%3D0%2Cnso-enfk%3D1%2Cnso-usnt%3D1%2Cnso-qnt-npqp%3D0-1633%2Cnso-qnt-npdq%3D0-5608%2Cnso-qnt-npt%3D0-1229%2Cnso-qnt-ndc%3D2051%2Ccspa-dspm-nm-mnp%3D0-06145%2Ccspa-dspm-nm-mxp%3D0-153625%2Cnso-unt-npqp%3D0-1506%2Cnso-unt-npdq%3D0-4694%2Cnso-unt-npt%3D0-061%2Cnso-unt-ndc%3D300%2Ccspa-uipm-nm-mnp%3D0-007625%2Ccspa-uipm-nm-mxp%3D0-053375...0...1.1.64.psy-ab..2.0.0.Gv5RYBfHmfU https://www.google.com/search?q=fast+object+detection+deep+l... and https://www.reddit.com/r/MachineLearning/search?q=object+detection&restrict_sr=on&sort=new&t=all https://www.reddit.com/r/MachineLearning/search?q=object+det... There's been an absolute deluge of papers, I can't say I've kept up with them all. There was one interesting one in particular was able to learn object detection in an unsupervised manner in a novel improvement upon Bottou's "is object detection for free" paper.
- anc84 9y ago> There's been an absolute deluge of papers, I can't say I've kept up with them all. That's why recommendations were asked for.
- maffydub 9y agoTo save someone googling, the referenced "Is object detection for free" paper is at http://leon.bottou.org/publications/pdf/cvpr-2015.pdf http://leon.bottou.org/publications/pdf/cvpr-2015.pdf I also just came across "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks", which looks interesting and is at https://arxiv.org/pdf/1506.01497.pdf https://arxiv.org/pdf/1506.01497.pdf (I'll be reading these after work tonight.)
- edshiro 9y agoNice! Btw this guy's CV is awesome :) : https://pjreddie.com/resume/ https://pjreddie.com/resume/
- ge96 9y agoWhoa that was a download link (On Android)
- kusha 9y agoChrome on Android auto downloads PDFs.
- ge96 9y agoOdd I'm in firefox. I think Chrome for me asked which app to open with but that probably means it was already downloaded. I don't know I've heard of malware with PDFs, I know probably a stretch on Android and this person/HN probably not really a concern. Still don't like auto-download unless expecting one. But I get it was the lack of a browser PDF displayer. Maybe he isn't running PDF.js or whatever that is you run to display pdfs on your site. (Sorry for so much text, TL;DR I'm afraid of what I don't know) I think I'm secure, but am I really secure?
- quickthrower2 9y agoI wouldn't hire someone who formats their CV like that. Employment is a serious proposition, and ... Ah just kidding. Nice CV.
- taneq 9y agoIt's the kind of thing that you can only get away with if you're very, very good. At that point it becomes a kind of power move, like showing up to a high end interview in jeans and a t-shirt.
- systemtest 9y ago
- giza182 9y agoImpressive. Does anyone actually know of real apps that make use of object detection? Apart from Google apps.
- shahar_m 9y agoThe tutorial says "Edit src/yolo.c, lines 54 and 55" to train for different classes, but I don't have any yolo.c file at src folder... any ideas? [https://github.com/pjreddie/darknet/wiki/YOLO:-Real-Time-Object-Detection https://github.com/pjreddie/darknet/wiki/YOLO:-Real-Time-Obj...]