WEBVTT 00:00.080 --> 00:03.240 We uploaded the images of the battery. 00:03.280 --> 00:10.800 Now let's label those images, click label myself creating job loading images. 00:11.040 --> 00:12.280 And now you see. 00:12.320 --> 00:16.720 Guys, this is a tip here to help you build a good model. 00:17.280 --> 00:25.720 This is not a good model because uh, the the annotation and the box is drawing around the head and 00:25.720 --> 00:30.960 the helmet while the correct one is drawing just on the helmet. 00:30.960 --> 00:37.960 So you see, guys, this is our annotation, um, annotation tip click get Started. 00:38.160 --> 00:40.800 And here let's start annotating. 00:40.840 --> 00:46.520 You see guys 63 images you can create uh, more okay. 00:46.560 --> 00:50.560 You see this is the smart select using Sam A3. 00:50.760 --> 00:52.120 Click finish. 00:52.240 --> 00:54.400 And you see guys this is the object. 00:54.440 --> 00:56.160 This is the battery. 00:56.200 --> 00:59.880 You can name it as battery okay. 01:00.160 --> 01:00.960 Save. 01:00.960 --> 01:02.440 And it's saved. 01:02.440 --> 01:04.800 So you see guys this is the first one. 01:05.080 --> 01:07.160 Now let's annotate the second one. 01:07.160 --> 01:10.480 You can draw the box like this drawing like this. 01:10.640 --> 01:15.860 Or you can select using the Sam three, click finish. 01:15.900 --> 01:21.100 You see guys this is the battery in uh in purple and yellow objects. 01:21.100 --> 01:25.820 You can delete the object, the entire object or the battery. 01:25.820 --> 01:28.060 So battery is not detected. 01:28.060 --> 01:31.420 So you need to uh to delete it. 01:31.460 --> 01:36.620 You can undo using this going to layers delete. 01:36.660 --> 01:40.940 And here select delete delete if you missed anything. 01:40.940 --> 01:48.420 If you uh if you annotate it by by mistake here we have use this smart Select. 01:48.420 --> 01:51.780 You can also use the bounding box tool. 01:51.780 --> 01:54.020 So I'm going to draw like this. 01:54.020 --> 01:57.780 This is the battery and I name it as battery. 01:58.020 --> 02:01.340 You can select this class battery okay. 02:01.660 --> 02:03.380 The image number three. 02:03.540 --> 02:04.860 This is the battery. 02:04.980 --> 02:05.940 Click save. 02:06.140 --> 02:14.780 So as you see guys there are a lot of tools that uh help you uh smart selecting the image. 02:14.940 --> 02:19.380 You see guys this is Sam three and this is the battery you see. 02:19.420 --> 02:27.400 Or you can use the standard bounding box or the polygon or the brush tool and other tools. 02:27.400 --> 02:30.080 So the most important thing is bounding box. 02:30.080 --> 02:35.520 Using the shortcut is B and the smart select is S okay. 02:35.720 --> 02:39.560 Bounding box I'm drawing drawing this battery save. 02:39.680 --> 02:48.400 Next this is the box and save next this is the battery and so on okay. 02:48.440 --> 02:56.000 So whenever you finish 63 images click on this Add images to data set. 02:56.000 --> 03:00.840 And here we go I labeled this all the images 63 images. 03:00.840 --> 03:06.960 You can see all the images are being labeled correctly using the bounding boxes. 03:06.960 --> 03:09.520 So add images to data set. 03:09.560 --> 03:16.000 Click okay and labeled images to add 63 remaining batch images. 03:16.000 --> 03:19.640 Zero are labeled every image. 03:19.640 --> 03:22.960 Here we have train validation and test. 03:23.000 --> 03:27.480 We need to split and make a distribution of our images. 03:27.480 --> 03:31.360 So here you can split images between train and test. 03:31.400 --> 03:32.800 Add all images to training. 03:32.800 --> 03:39.050 Set all images to validate the validation set and all images to testing set. 03:39.090 --> 03:47.010 You can, um, use existing values or split images between training, validation and test. 03:47.210 --> 03:54.770 70% for training is good, valid 20% and test 10% is good. 03:54.810 --> 04:01.130 We have to train 44 images valid 613 images and test six images. 04:01.130 --> 04:09.290 All those values are very important because we need to get a 70% for training, 20% for validation. 04:09.290 --> 04:17.690 You can increase the number for, for example, 5,075% for training and validation, 15% and testing. 04:17.690 --> 04:18.530 This is good. 04:18.570 --> 04:22.770 Add 63 images and congratulations guys. 04:22.770 --> 04:25.810 This is our data set okay. 04:25.970 --> 04:27.770 So this is the data set. 04:27.810 --> 04:30.010 We can select the train. 04:30.290 --> 04:37.650 Those are the images that are selected for training those images for validation and those images for 04:37.810 --> 04:38.970 testing okay. 04:39.170 --> 04:44.490 So select all and here is our data set.