YOLOv7 Segmentation | Concrete Crack Detection | Google Colab | step-by-step Tutorial

[@aminemami3549]

On point, as always, Piotr 👌

[@SkalskiP]

Thank you very much! I'm doing my best :)

[@kushshrestha8722]

This guy seriously needs a pay rise. Awesome content

[@SkalskiP]

It's Peter from the video :) Thanks! I'll pass that idea to my superiors haha

[@vishalpahuja2967]

Hi can you show annotation process for cracks on a single wall and detecting. Thank you.

[@Roboflow]

I used a model that was already on Roboflow Universe. I did not annotate it myself.

[@vishalpahuja2967]

@@Roboflow Thankyou for the response , I actually needed to detect lines similar to cracks and wanted to know how can it be annotated so that detection can show exact shape of lines which can be curved so, how can I annotate the images and train the model to detect it.

[@Roboflow]

@@vishalpahuja2967 ah the data itself was annotatwed using polygons.

[@ksteven4469]

Thank you for your tutorial. I have a question for you. At around 2:55, you mentioned that you had previously completed a project that involved object detection and instance segmentation at the same time. Would it be possible for me to take a look at the code for that project? Thx

[@Roboflow]

Hi sorry for the late reply I was working on a new video. I did detection + pose estimation. Take a look here: github.com/SkalskiP/sport/blob/master/football-players-pose-estimation/football_players_pose_estimation.ipynb

[@robosergTV]

Thanks, great vid! YOLO7 creates an image with predictions during training, as you've shown it. Is it a feature of YOLO7 codebase, since YOLO5 does not generate an image with predictions automatically for me, only YOLO7 codebase does (as in your video)

[@SkalskiP]

Hi 👋! It's Peter from video. Doesn't it? I'm pretty sure it does. I'd need to double check in that case.

[@BenjaminvonCramon]

Is it hard fact that Roboflow only accepts square format? I'd so prefer obviating the need to subdivide 1.5 aspect rasters into square tiles.

[@vikashkumar-cr7ee]

Dear Tutor, Greetings! I downloaded the Yolo format of the dataset. In the train folder, I can see images and corresponding labels folder, But I can't see any the yellow label that is created on crack. Infact, this is shown in the tutorial, but in the actual dataset, it is missing. It seems like it is not needed as a label is already done.

[@joshuamacasadia4495]

How can i deploy the model to roboflow?

[@muhammadmuzammul1023]

Great knowledge

[@abangfikri1865]

thank u for the video, it helped me alot. However is yolov5 or yolov7 segmentation model can be deployed on android? is it possible? and how can it be done?

[@SkalskiP]

Hi, its Peter from the video. I'm not sure about YOLOv7 as I know some of their exports doesn't work. However, you can for sure export YOLOv5 to TensorFlow Light and in that format it should be runnable on Android device.

[@abangfikri1865]

@@SkalskiP ohh thank u very for the reply. For yolov5 i know it can be runnable on android for object detection model. But can it be done for yolov5 segmentation?

[@serchengxu4399]

can you do a real time crack detection with using yolov7 ?

[@Roboflow]

Realistically speaking I doubt that. We have really long backlog. However I encourage you to use this video and our other video about real time video processing. You should be able to figure it out ;) good luck 🍀

[@janithaprathapa1267]

how to deploy this model in local machine

[@Laddu225]

Hey. I think it's semantic segmentation not instance segmentation.

[@SkalskiP]

Hi it's Peter from the video. It is actually instance segmentation. On the results image around 0:10 you can see we have multiple individual detections per image, not just single mask.

[@christian.js.1997]

Thanks for this tutorial, I've been searching for hours on how to display/visualize feature map in YOLOv7, please make a tutorial about that. 😁

[@Roboflow]

You would like to visualize stored in each layer of the network?

[@christian.js.1997]

@@Roboflow Yes 🤔

[@angelospapadopoulos7679]

is it possible and how can we train from scratch here?

[@afiedoh6228]

Hello, Thank you for your video. Please how do i apply my best.pt file to realtime video from my webcam? Thank you

[@Roboflow]

Hi! Please ask the question in Notebooks Repository: github.com/roboflow/notebooks/discussions/categories/q-a we'll try to help you :)

[@afiedoh6228]

@@Roboflow Thank you, just figured it out!

[@Roboflow]

@@afiedoh6228 great to hear that :)

[@youssefkhaled5331]

@@afiedoh6228 could you pls tell us how

[@nabiladnan626]

when start train: AttributeError: module 'numpy' has no attribute 'int'.

[@elbadamohamed6605]

Hi, Hi did you find the solution? i'm struggling with it

[@nabiladnan626]

@@elbadamohamed6605 nope. didn't try later. still its happening ?

[@nabiladnan626]

@@elbadamohamed6605 no solution still. did u find anything ?

[@nabiladnan626]

@@elbadamohamed6605 still stuck. Did u find any way?

[@nabiladnan626]

​@@elbadamohamed6605still no solution.. Did you find anything to fix?

[@dinanmutamaddin400]

how if i want to change thickness bonding box and size of font when predict object

[@Roboflow]

Use supervision it is a lot more flexible when it comes to annotating detections.

[@YapHokLai]

how to save the model?

[@Roboflow]

It is already saved after training and stored in the runs directory.

[@iceiceisaac]

Good career move lol

[@Roboflow]

Haha what do you mean?

[@iceiceisaac]

@@Roboflow I’m a civil engineer too ! Really thinking about switching but I’m too new to this stuff

[@rishabhsheoran6959]

Hey! can you please help me with the deployment of the yolov7 segmnetation model on triton? When I hit the triton inference server, I get back following outputs: name: output tensor: float32[batch,anchors,Concatoutput_dim_2] name: onnx::Slice_539 tensor: float32[Transposeonnx::Slice_539_dim_0,3,Transposeonnx::Slice_539_dim_2,Transposeonnx::Slice_539_dim_3,40] name: onnx::Slice_693 tensor: float32[Transposeonnx::Slice_693_dim_0,3,Transposeonnx::Slice_693_dim_2,Transposeonnx::Slice_693_dim_3,40] name: onnx::Slice_844 tensor: float32[Transposeonnx::Slice_844_dim_0,3,Transposeonnx::Slice_844_dim_2,Transposeonnx::Slice_844_dim_3,40] name: 517 tensor: float32[Mul517_dim_0,32,Mul517_dim_2,Mul517_dim_3] output (1, 100800, 40) onnx::Slice_539 (1, 3, 160, 160, 40) onnx::Slice_693 (1, 3, 80, 80, 40) onnx::Slice_844 (1, 3, 40, 40, 40) From the above outputs how do i extract the bounding boxes and the masks?