DETECTRON2 Custom Object Detection, Custom Instance Segmentation: Part 2 (Training Custom Models)

[@TheCodingBug]

► My Other Tutorials: DETECTRON2 Custom Object Detection, Custom Instance Segmentation Tutorial: Part 2 (Training Models) kzbin.info/www/bejne/faCspauoZpuUm5o DETECTRON2 Custom Object Detection, Custom Instance Segmentation Tutorial: Part 1 (Preparing Data) kzbin.info/www/bejne/nJe3hoV3Za-AZ7M Detectron2 on Colab kzbin.info/www/bejne/mJvEqmqciZ5mars Instance Segmentation as Rendering kzbin.info/www/bejne/nZ-7nXhvd7RjotE Detectron2 Complete Tutorial kzbin.info/www/bejne/hpOWoKN7e7Vsarc Colorize Black and White Images and Videos using Python OpenCV kzbin.info/www/bejne/e4u6eXSaZa57Z68 Build and Install OpenCV 4.5.1 With CUDA GPU Support on Windows 10 kzbin.info/www/bejne/j6TQmX6Xp52ihcU Face Detection Using OpenCV Python with CUDA GPU Acceleration kzbin.info/www/bejne/fYnGqmp6npair9U YOLOv4 On Android Using TFLite kzbin.info/www/bejne/j6ukm3SJa7SmbsU Install TensorFlow GPU Under 90 Seconds kzbin.info/www/bejne/qqCtlmuQl6ube8U Install PyTorch GPU Under 90 Seconds kzbin.info/www/bejne/qJKlnJuLpMqCftU Custom YOLOv4 Object Detection with TensorFlow and TFLite kzbin.info/www/bejne/rKu3dH2DZp2opa8 Darknet YOLOv4 Custom Object Detection: Part 2 (Training YOLOv4 Darknet) kzbin.info/www/bejne/Y3-oc2iGYrh_n5I Darknet YOLOv4 Custom Object Detection: Part 1 (Preparing Custom Dataset) kzbin.info/www/bejne/qXynqqaEqdObgJY YOLOv4 Object Detection with TensorFlow, TFLite and TensorRT kzbin.info/www/bejne/qnTQdGqkrst9ppo Darknet YOLOv4 Object Detection for Windows 10 on Images, Videos, and Webcams kzbin.info/www/bejne/fHaVeHWbgNqkpsU Real Time Object Detection on Webcam and Videos Using OpenCV With YOLOv3 and YOLOv4 | Windows Linux kzbin.info/www/bejne/fJvcd2NpqsaqqaM Build and Install OpenCV 4.4.0 with CUDA (GPU) Support on Windows 10 kzbin.info/www/bejne/qpu7nIpmYpmag6c Install TensorFlow GPU and PyTorch with CUDA on Windows 10 Anaconda | CUDA 10.1 cuDNN 7.6 kzbin.info/www/bejne/hp26aq13nJmWrpo Real-time Multiple Object Tracking with YOLOv4 TensorFlow and Deep Sort | Linux, Windows kzbin.info/www/bejne/fZLKenxrhaaYgck

[@coder5199]

Hi thanks for this video. There are TONS of YT vids on Detectron2 on object detection and instance segmentation but NONE on semantic segmentation for custom datasets. Can you be the one that fill the gap?

[@yhnshkn]

Thanks for this tutorial!!! Was trying to implement this on VS code instead of Google Colab and now I can finally do it! My next step is on how to deploy it to Raspberry Pi. If it is possible, hope you can create a video of it soon. Thanks once again

[@gustavosantiago4514]

Thank you so much!! After almost a week looking for a good tutorial, I finally found one!

[@kiphaynes]

Hello, Thank you again for this tutorial. I am having the same warning that you are after training: d2.engine.defaults]: No evaluator found. Use `DefaultTrainer.test(evaluators=)`, or implement its `build_evaluator` method. I do not believe the model is doing any testing after training. Suggestions on this?

[@gauravsingh-sd3hw]

Wonderful video!!! Can you please make a detailed video on "Panoptic segmentation using Detectron2"?

[@TheCodingBug]

If you are looking for pretrained Panoptic Segmentation model using Detectron2, it's already covered here: kzbin.info/www/bejne/hpOWoKN7e7Vsarc But none for custom dataset yet.

[@otavioaugustomedolaconquis598]

Nicely done! Loved the video. One question: do I need to have internet connection once my model is trained? I mean, I have to train the Detectron first, but once I've done this step, will the code run and identify my images without internet connection? Greetings from Brazil!

[@TheCodingBug]

Yes it will. You just need internet connection for the first time the model is downloaded.

[@gauravsingh-sd3hw]

Yeah!!! That's with any Model. You download weights and all dependencies locally once and thereafter you don't need any internet.

[@grantrichardet8521]

THANK YOU SO MUCH! I loved the video. My only question is: if we are doing object detection or instance segmentation on training on a set with more than one class. How do we get the labels to appear as more than just a number when we test the model?.

[@fernandolee7039]

Great tutorial! I tried with my own dataset but I got this error (TypeError: 'NoneType' object is not subscriptable) when I run this (plot_samples(dataset_name=train_dataset_name, n=2)). Not sure why?

[@jasonbourn29]

Can you do a video on detection and tracking(sort ,deepsort,strong sort)with yolov8

[@SebastianBejas]

Great tutorial! Everything is working like a charm except for the fact the video is extremely slow, why would that be?

[@TheCodingBug]

Check if you're able to use GPU.

[@au-yeungwaikwong3846]

hi there, thanks for the video and it is very useful. in case if we trained the model one time and want to continue improving the performance, how we can train the model again and again with additional data based on the previous trained model. thanks

[@ilyassziz5]

hy, thank you for this tutoriel sir i followed all steps but i get error : cannot find field '_fields'in the given instances

[@陽明交大-高明秀]

Great! can you explain how to count number of parameter in a model? Thanks

[@ashutoshkumawat2935]

Amazing Tutorial

[@aboudezoa]

Very nice detailed tutorial 👍🏻

[@TheCodingBug]

I am glad you found it helpful.

[@JoelPrabhod]

Nicely done! how do we calculate the area of each mask from the inference?

[@GeoHan808]

also curious about that

[@UZMAALFATMI]

Thanks a lot for this tutorial. Just one thing - Please dont include the background music, its disturbing.

[@TheCodingBug]

I avoid using it in new tutorials.

[@tazuddin5831]

Thanks a lot for the video. How to put the class name on box after object detection ?

[@franciscoalberto4774]

If I want the network to be re-trainable, that is, train the model with the dataset I have and then when I have more images, continue training it. For that in the step where you put trainer.resume_or_load(resume=False) I would have to put resume=True??

[@yangzhaodong]

一个非常好的视频，非常感谢。我是初学者。请问是否有人知道如何保存正在运行的视频

[@muddassarjilani4637]

excellent video, thank u sir

[@ShravanKumar147]

What do you prefer for object detection - Yolo or Detectron2? How do you compare, and what factors help in the decision?

[@TheCodingBug]

YOLOv7 because of its speed and accuracy. I go for official paper for comparisons.

[@omarbaruch5328]

I found this problem by following all your stpes, one by one [Errno 2] No such file or directory: 'datasets\\coco/annotations/instances_train2017.json'

[@alperenclk]

Sir I love your videos. But i have a problem "No valid data found in my_dataset_train." please help me

[@leandromontenegropinto6704]

Very Good Video. Thanks =)

[@franciscoalberto4774]

Hello, when execute the function plot_sample tells me TypeError: string indices must be integers. Why? Help me pls! Copy the same what in video

[@damienbrocard940]

Hello ! thanks dor this tutorial. Do you know how i can add labels when i do test.py ?

[@kakolibora6937]

Is detectron2 can handle only annotations with even number of points for a class? I mean to say if an object with a specific class is annotated with 8 points, then all other objects of that class should have only 8 points.

[@TheCodingBug]

No. Any number of points would work

[@muddassarjilani4637]

Thanks a lot, Sir, can we apply this model using flask

[@TheCodingBug]

Yes you can.

[@muddassarjilani4637]

@@TheCodingBug Respected Sir, kindly make a short video, if you have time. Your effort will be highly appreciated.

[@muddassarjilani4637]

sir, how we can deploy this model using a Streamlit app, kindly make a video on this like this

[@yahaisha]

tq so much sir👏

[@thelazydeveloper]

what if i want to extract the mask white inside black beacause i want to replace it with some thing how should i proced

[@TheCodingBug]

in predictions returned by default predictor, you can find the bounding box values in "pred_boxes" key and the segmentation masks in "pred_masks" key. Masks are returned in True False format (1,0).

[@thelazydeveloper]

@@TheCodingBug thanks this was so helpfull

[@thelazydeveloper]

will this work with multiple objects 3 for example

[@TheCodingBug]

Yes it would.

[@deidy2005]

Man, I created a customized model with 3 classes (A, B and C). And I would like to take especific masks for a especific object. And I use this code: mask_A = outputs['instances'].pred_masks.cpu().numpy()[0]

[@mannversteckter5631]

thanks very much for this tutorials!!!

[@TheCodingBug]

I'm glad you found it helpful.

[@manujkumarjoshi9342]

why we annotate test images??

[@TheCodingBug]

To get metrics at testing times by comparing predictions vs actual labels.