Detectron2 - Next Gen Object Detection Library - Yuxin Wu

[@magelauditore333]

I can go on hearing it for days. Wish someone had told me about the world of ML,DL earlier. Feeling like wasted my 4 year of engineering

[@Anonymous-vh9tc]

At the same time it is also the most OVERHYPED field. A good engineer can learn to take advantage of ML in actual practical applications. And no you don’t need a PhD I math.

[@anirudhsilverking5761]

@@Anonymous-vh9tc If you need a good job, I guess you do

[@dennismertens990]

@@anirudhsilverking5761 you do need math to understand the algorithms. Otherwise, what kind of mediocre engineer would you be? Using algorithms without even having a minimum idea of how they work.

[@SuperOnlyP]

You are not wasting your time. AI is just a tool to combine with other tools.

[@robosergTV]

I mean if for the last 10 years you didnt hear about ML/AI or Deep Learning you must have been living under a rock. Even regular news outlets write about it (often in context of self driving cars etc)

[@MinhTran-ew3on]

love the way you answer github issues, ppwwyyxx

[@cahanne9411]

Can detectron2 be used to classify 3D point clouds Or is it only for 2D images?

[@bytblaster]

2D only. But you can map your 3D point cloud to multiple 2D images

[@JOHN-vb5bh]

Is there anyone who were able to do validation in detectron2 between the training

[@gluxodin3523]

when will you add cuda11 and RTX 3000 series Compatability???? Can't wait for using my new RTX 3090.

[@PyTorch]

Hi Glux, PyTorch 1.7 works with CUDA 11 `conda install pytorch torchvision torchaudio cudatoolkit=11.0 -c pytorch` or `pip install torch==1.7.1+cu110 torchvision==0.8.2+cu110 torchaudio===0.7.2 -f download.pytorch.org/whl/torch_stable.html`

[@ahmadshah661]

any news on nash artheon????? guys i need info on that .. cant find any. help! :(

[@somyarani9250]

Can I use YOLO model trained on customized data with detectron2 ?

[@PyTorch]

Hi Somya. For help, please join and post in the PyTorch Forums: discuss.pytorch.org

[@emilbogomolov5709]

Yes, the relevant question is: could we build one-stage detector using detectron2 without changing the source code?

[@Ruhgtfo]

Dayum

[@anji.reddy.akkala]

could you please explain how to use new class with pre-trained classes? in future, if i want to add few more classes to trained model, how to do the same?

[@PyTorch]

Hi Anjireddy. For help, please join and post in the PyTorch Forums: discuss.pytorch.org

[@anirudhsilverking5761]

5:14, those were direct shots at tensorflow OD API.

[@vincentnaayem]

Do a updatable package!!!!! Please!

[@crytoy87]

Can Detectron2 be ran on Raspberry pi ?

[@PyTorch]

Hi Crytoy, for further assistance please join and post in the PyTorch Forums: discuss.pytorch.org

[@deepakraj008]

Ya it can run.

[@Venkateshtanniru9]

I'm looking for more information about detectron2, if you could help me with that it helps me a lot for my project

[@PyTorch]

Hi Tanniru, you can find more information on Detectron2 here: bit.ly/2q1gPRP

[@harutmargaryan9980]

Can I use Detectron2 with SSD algorithm?

[@PyTorch]

Hi Harut. For help, please join and post in the PyTorch Forums: discuss.pytorch.org

[@JOHN-vb5bh]

Can anyone give suggestions or instructions on integrating this code with detectron2 official code. I had done from detectron2.engine import MyTrainer from detectron2.engine import LossEvalHook and trainer = MyTrainer(cfg) in my training scripts But getting an error TypeError Traceback (most recent call last) in 21 22 os.makedirs(cfg.OUTPUT_DIR, exist_ok=True) ---> 23 trainer = MyTrainer(cfg) 24 trainer.resume_or_load(resume=False) 25 trainer.train() TypeError: 'module' object is not callable Please help me to solve this problem