WEBVTT 00:00.080 --> 00:02.360 Hello developers, and welcome back. 00:02.400 --> 00:09.800 In this video, we're going to learn about Roboflow and why we needed to build our custom model. 00:09.840 --> 00:18.760 Roboflow is an end to end platform that simplifies and accelerates the process of building and deploying 00:18.760 --> 00:21.600 computer vision models. 00:21.760 --> 00:30.880 It provides a suite of tools that handle the entire workflow, from preparing your images to deploying 00:30.920 --> 00:35.080 a trained model, all with a user friendly interface. 00:35.200 --> 00:41.680 Building a computer vision model, like an object detection or image classification, is traditionally 00:41.720 --> 00:45.720 complex and requires deep expertise. 00:46.080 --> 00:52.960 The most time consuming and tedious part is in the model training itself, but the data preparation. 00:53.000 --> 00:59.400 Roboflow automates and streamlines this data preparation stage. 00:59.440 --> 01:09.400 Okay, so we're going to use the roboflow, um, in in many fields you can make an automotive, telecommunication, 01:09.640 --> 01:15.760 healthcare, retail, banking, agriculture, utilities, government and other ways. 01:15.760 --> 01:23.400 So this is a very important website we're going to use to build our custom model okay. 01:23.640 --> 01:30.800 Go to roboflow comm, get sign in and get started. 01:30.800 --> 01:34.200 Sign in with Google or any account. 01:34.240 --> 01:36.800 Enter the name of your workspace. 01:36.800 --> 01:42.160 I'm gonna name it as Object Detection and make it for free. 01:42.200 --> 01:43.400 Free forever. 01:43.400 --> 01:44.840 Because it's public plan. 01:44.840 --> 01:50.000 You can buy the premium trial, Create Workspace, and here we go. 01:50.120 --> 02:02.640 Now let's go to the project Create New Project and select Object Detection and name the project as Object 02:02.680 --> 02:03.680 detection. 02:03.880 --> 02:04.840 Android. 02:05.000 --> 02:12.830 The annotation groups objects, make it the public and keep this CC by 4.0. 02:13.030 --> 02:13.630 Okay. 02:13.670 --> 02:15.470 Click continue with public. 02:15.510 --> 02:18.190 We don't need the the rapid. 02:18.230 --> 02:21.150 Use traditional model builder instead. 02:21.430 --> 02:23.750 And wait for creating the project. 02:23.790 --> 02:24.470 Here we go. 02:24.830 --> 02:27.190 Now we need to upload the data. 02:27.390 --> 02:36.710 The data is, uh, is composed and consists of images that we're going to annotate and tell our, our 02:36.710 --> 02:39.910 model that those are the detected objects. 02:39.950 --> 02:43.630 Now select any object that you want. 02:43.670 --> 02:47.030 You can select files from your computer. 02:47.190 --> 02:52.510 Or you can select folder directly with uh with many formats. 02:52.630 --> 03:02.950 You can upload images jpg, png, BMP, WebP, and the annotations in 26 formats. 03:03.110 --> 03:08.350 The videos in MOV and mp4 and the PDFs dot PDF. 03:08.390 --> 03:12.150 Okay, there are a lot of things you can upload here. 03:12.270 --> 03:15.230 Also, you can upload data from your phone. 03:15.230 --> 03:21.310 Scan the QR code to upload images and videos from your phone directly to your project. 03:21.310 --> 03:25.830 Also, you can search the Roboflow universe and bulk upload images. 03:26.070 --> 03:32.510 I'm gonna use for for making fast and rapid, uh, rapid detections. 03:32.510 --> 03:40.190 I'm gonna use my mobile phone to detect the objects and capture the images for later annotation and 03:40.190 --> 03:40.830 labeling. 03:40.910 --> 03:49.990 I change it to, um, to to light mode, and I'll scan the QR code on my phone. 03:50.030 --> 03:50.510 Okay. 03:50.710 --> 03:52.470 And open the browser. 03:52.670 --> 03:53.950 This is my phone. 03:53.950 --> 03:57.830 You can choose the files from your phone or start camera. 03:57.990 --> 03:59.590 I'll start the camera. 03:59.590 --> 04:00.550 And here we go. 04:00.830 --> 04:05.590 I, I need to, uh, you can select whatever, any object. 04:05.590 --> 04:08.110 I'm gonna select this battery. 04:08.110 --> 04:12.870 So this is our battery that we're going to detect okay. 04:12.950 --> 04:17.070 So start capturing the pictures you see uploading. 04:17.270 --> 04:20.820 And here to the left you see that? 04:20.820 --> 04:22.940 We uploaded the first picture. 04:23.140 --> 04:24.860 The second picture. 04:25.060 --> 04:26.260 Third picture. 04:26.540 --> 04:27.340 You see. 04:27.380 --> 04:33.340 Now start taking photos to the to the battery and of the battery. 04:33.380 --> 04:34.940 And like this. 04:35.220 --> 04:36.260 Like this. 04:36.420 --> 04:39.260 And with different angles, guys. 04:39.260 --> 04:47.020 Because we need to, to use it in different formats, in different shapes, in different contrasts. 04:47.260 --> 04:51.940 So this is very very very important to take pictures. 04:51.940 --> 04:59.900 So as you take in different in different situations your object detection model will be better. 04:59.940 --> 05:06.140 Also you can in different backgrounds in different different lighting okay. 05:06.340 --> 05:09.620 So this is very important. 05:09.620 --> 05:10.540 And here we go. 05:10.620 --> 05:19.540 When you finish uploading your images all those images are ready to label and and start telling Roboflow 05:19.580 --> 05:24.260 that the the object is this battery that we need to detect.