title
Determining other vehicle distances & collision warning (object detection) - Self Driving Cars p.18

description
Hello and welcome to another self-driving cars tutorial, in this tutorial we're going to use the TensorFlow Object Detection API to determine whether or not other vehicles on the road are too close. Text tutorials and sample code: https://pythonprogramming.net/detecting-distances-self-driving-car/ https://twitter.com/sentdex https://www.facebook.com/pythonprogramming.net/ https://plus.google.com/+sentdex

detail
{'title': 'Determining other vehicle distances & collision warning (object detection) - Self Driving Cars p.18', 'heatmap': [{'end': 1129.492, 'start': 1113.584, 'weight': 1}], 'summary': 'Using tensorflow object detection api, the video series focuses on detecting and determining distances of cars in grand theft auto, optimizing vram usage, measuring object distance, precision calculation, and car detection for self-driving cars.', 'chapters': [{'end': 47.128, 'segs': [{'end': 47.128, 'src': 'embed', 'start': 2.538, 'weight': 0, 'content': [{'end': 5.699, 'text': 'what is going on, subscribers and others.', 'start': 2.538, 'duration': 3.161}, {'end': 13.541, 'text': 'uh, welcome to the second video of object detection inside of grand theft auto using the tensorflow object detection api.', 'start': 5.699, 'duration': 7.842}, {'end': 16.602, 'text': "in this video what we're going to do is build on the last video.", 'start': 13.541, 'duration': 3.061}, {'end': 26.165, 'text': "in the last video we just got object detection working in grand theft auto and now what we're going to go ahead and do is try to detect cars in the distance basically,", 'start': 16.602, 'duration': 9.563}, {'end': 31.987, 'text': 'and then try to figure out some way to calculate their distance from us,', 'start': 26.165, 'duration': 5.822}, {'end': 35.328, 'text': "like are they far enough away that they're not a problem or are they really close?", 'start': 31.987, 'duration': 3.341}, {'end': 38.529, 'text': 'and then from there we can build on top of that to, uh,', 'start': 35.328, 'duration': 3.201}, {'end': 45.292, 'text': 'detect when a vehicle maybe is too close and either take evasive action or just display like a warning or something like that.', 'start': 38.529, 'duration': 6.763}, {'end': 47.128, 'text': "so That's what we're going to be doing now.", 'start': 45.292, 'duration': 1.836}], 'summary': 'Developing car distance detection in gta using tensorflow object detection api.', 'duration': 44.59, 'max_score': 2.538, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA2538.jpg'}], 'start': 2.538, 'title': 'Object detection in grand theft auto', 'summary': 'Focuses on using the tensorflow object detection api to detect cars in the distance in grand theft auto and determining their proximity to take appropriate actions.', 'chapters': [{'end': 47.128, 'start': 2.538, 'title': 'Object detection in grand theft auto', 'summary': 'Focuses on building upon the previous video to detect cars in the distance in grand theft auto using the tensorflow object detection api and determining their proximity to take appropriate actions.', 'duration': 44.59, 'highlights': ['The video aims to enhance object detection in Grand Theft Auto by detecting cars in the distance and calculating their proximity to take necessary actions.', 'The last video focused on getting object detection working in Grand Theft Auto.', 'The goal is to develop a system that can detect when a vehicle is too close and then take evasive action or display a warning.', "The chapter aims to calculate the distance of detected cars from the player's vehicle to determine if they are far enough to not pose a problem."]}], 'duration': 44.59, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA2538.jpg', 'highlights': ['The video aims to enhance object detection in Grand Theft Auto by detecting cars in the distance and calculating their proximity to take necessary actions.', 'The goal is to develop a system that can detect when a vehicle is too close and then take evasive action or display a warning.', "The chapter aims to calculate the distance of detected cars from the player's vehicle to determine if they are far enough to not pose a problem.", 'The last video focused on getting object detection working in Grand Theft Auto.']}, {'end': 205.967, 'segs': [{'end': 74.184, 'src': 'embed', 'start': 47.588, 'weight': 0, 'content': [{'end': 52.53, 'text': 'From the last video, since I was recording and noticing frame drops, I went ahead and just kind of changed something.', 'start': 47.588, 'duration': 4.942}, {'end': 65.695, 'text': 'In here, you can designate how much VRAM the session wants to take up here by adding these GPU options.', 'start': 53.97, 'duration': 11.725}, {'end': 72.261, 'text': 'And then when you run your actual session in the config parameter there, you can add in some configuration information.', 'start': 65.715, 'duration': 6.546}, {'end': 74.184, 'text': "So I'm gonna try to set it lower.", 'start': 72.301, 'duration': 1.883}], 'summary': 'Adjusted vram settings to reduce frame drops during recording.', 'duration': 26.596, 'max_score': 47.588, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA47588.jpg'}, {'end': 152.235, 'src': 'embed', 'start': 126.207, 'weight': 1, 'content': [{'end': 133.632, 'text': "so basically, what we're going to do is, before the m show, we're just going to work right underneath the vizutil dot,", 'start': 126.207, 'duration': 7.425}, {'end': 140.153, 'text': "visualize boxes and labels and all that, And we're basically going to start iterating through.", 'start': 133.632, 'duration': 6.521}, {'end': 142.553, 'text': 'any one of these would work.', 'start': 140.153, 'duration': 2.4}, {'end': 143.933, 'text': "I'm just going to iterate through boxes.", 'start': 142.593, 'duration': 1.34}, {'end': 147.694, 'text': "I am sure there's a better way to do all of the things I'm about to do.", 'start': 144.334, 'duration': 3.36}, {'end': 152.235, 'text': 'This is just an example, a proof of concept.', 'start': 148.294, 'duration': 3.941}], 'summary': 'Preparing to visualize boxes and labels before the show, iterating through them for better performance and proof of concept.', 'duration': 26.028, 'max_score': 126.207, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA126207.jpg'}], 'start': 47.588, 'title': 'Optimizing vram usage and object detection implementation', 'summary': 'Discusses optimizing vram usage by adjusting gpu options and implementing object detection in tensorflow, including iterating through boxes and visualizing the results.', 'chapters': [{'end': 205.967, 'start': 47.588, 'title': 'Optimizing vram usage and object detection implementation', 'summary': 'Discusses optimizing vram usage by adjusting gpu options, and implementing object detection in tensorflow by iterating through boxes and visualizing the results.', 'duration': 158.379, 'highlights': ['The chapter discusses optimizing VRAM usage by designating the amount of VRAM the session wants to take up by adding GPU options, potentially setting it to lower percentages like 50% or 70%.', 'In object detection implementation, the discussion focuses on iterating through boxes, checking classes, and visualizing the results using the vizutil function, serving as a proof of concept for further development and experimentation.']}], 'duration': 158.379, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA47588.jpg', 'highlights': ['The chapter discusses optimizing VRAM usage by designating the amount of VRAM the session wants to take up by adding GPU options, potentially setting it to lower percentages like 50% or 70%.', 'In object detection implementation, the discussion focuses on iterating through boxes, checking classes, and visualizing the results using the vizutil function, serving as a proof of concept for further development and experimentation.']}, {'end': 404.626, 'segs': [{'end': 234.107, 'src': 'embed', 'start': 205.967, 'weight': 0, 'content': [{'end': 208.389, 'text': "let's say, you've got a flat 2d image right?", 'start': 205.967, 'duration': 2.422}, {'end': 212.512, 'text': 'How could you detect the distance of an object in that image?', 'start': 208.789, 'duration': 3.723}, {'end': 221.023, 'text': 'Um, At least for me, I believe, the only way you could possibly do that is if you knew the size of that object beforehand.', 'start': 213.293, 'duration': 7.73}, {'end': 229.826, 'text': "So you can detect distances like, so for example, the average car, I don't know, three feet, four feet wide? I really don't know.", 'start': 221.723, 'duration': 8.103}, {'end': 231.946, 'text': "Let's go with like four feet wide.", 'start': 230.546, 'duration': 1.4}, {'end': 234.107, 'text': 'Maybe an SUV might even be five feet or something.', 'start': 232.006, 'duration': 2.101}], 'summary': 'Detecting distance of objects in 2d image based on known sizes.', 'duration': 28.14, 'max_score': 205.967, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA205967.jpg'}, {'end': 278.769, 'src': 'embed', 'start': 252.918, 'weight': 1, 'content': [{'end': 258.041, 'text': "Unfortunately, we don't know the exact size of every single car in Grand Theft Auto, but we don't need to know.", 'start': 252.918, 'duration': 5.123}, {'end': 262.424, 'text': 'We just need to kind of have a general idea like, okay, these objects are about these sizes.', 'start': 258.081, 'duration': 4.343}, {'end': 263.265, 'text': 'So for example.', 'start': 262.484, 'duration': 0.781}, {'end': 265.265, 'text': 'for a car.', 'start': 264.085, 'duration': 1.18}, {'end': 269.286, 'text': "what we're going to do is we're just going to basically measure the amount of pixels in between.", 'start': 265.265, 'duration': 4.021}, {'end': 273.067, 'text': 'you know from x, x1 to x2, how wide is that?', 'start': 269.286, 'duration': 3.781}, {'end': 278.769, 'text': 'and then from there we can get a relative determination of how far away that object actually is.', 'start': 273.067, 'duration': 5.702}], 'summary': 'Estimating car sizes in grand theft auto using pixel measurements for relative determination of distance.', 'duration': 25.851, 'max_score': 252.918, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA252918.jpg'}, {'end': 381.507, 'src': 'embed', 'start': 349.793, 'weight': 2, 'content': [{'end': 351.696, 'text': "We'll detect probably three things here.", 'start': 349.793, 'duration': 1.903}, {'end': 354.68, 'text': 'And even this might not work very well.', 'start': 352.336, 'duration': 2.344}, {'end': 360.027, 'text': 'But anyway, three, six, or eight.', 'start': 354.74, 'duration': 5.287}, {'end': 364.853, 'text': 'So a car is three, six is a bus, and eight is a truck.', 'start': 360.648, 'duration': 4.205}, {'end': 366.883, 'text': 'And you might be like syntax.', 'start': 364.954, 'duration': 1.929}, {'end': 368.584, 'text': 'how do you know that?', 'start': 366.883, 'duration': 1.701}, {'end': 377.846, 'text': 'so the way I did that, or I found that out, is let me just navigate there inside of the object detection models.', 'start': 368.584, 'duration': 9.262}, {'end': 380.047, 'text': "yes, I'm trying to think.", 'start': 377.846, 'duration': 2.201}, {'end': 381.507, 'text': 'was it in training?', 'start': 380.047, 'duration': 1.46}], 'summary': 'Object detection can identify three classes: car, bus, and truck.', 'duration': 31.714, 'max_score': 349.793, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA349793.jpg'}], 'start': 205.967, 'title': '2d object distance detection', 'summary': 'Covers methods for detecting object distance in 2d images, including calculating exact distance based on known object sizes, as well as using pixel measurements to estimate the size and distance of objects in gta, and the challenges associated with object detection and classification.', 'chapters': [{'end': 252.418, 'start': 205.967, 'title': 'Detecting object distance in 2d images', 'summary': 'Discusses the method of detecting object distance in 2d images by calculating the exact distance based on the known size of the object, such as in the case of detecting a car with a width between 4 and 5 feet.', 'duration': 46.451, 'highlights': ['Calculating the exact distance of an object in a 2D image is possible if the size of the object is known, such as in the case of detecting a car with a width between 4 and 5 feet.', 'The method involves using the known size of the object to calculate the distance based on its width in the image.', 'Understanding the size of the object beforehand is crucial for accurately determining the distance in a 2D image.']}, {'end': 310.32, 'start': 252.918, 'title': 'Vehicle size detection in gta', 'summary': 'Discusses using pixel measurements to estimate the size and distance of objects in grand theft auto, such as cars and pedestrians, to assess threats based on the percentage of screen pixels they occupy.', 'duration': 57.402, 'highlights': ['Using pixel measurements to estimate the size and distance of objects in Grand Theft Auto The method involves measuring the amount of pixels between specific points to determine the relative size and distance of objects, such as cars and pedestrians.', 'Assessing threats based on the percentage of screen pixels occupied by objects By setting thresholds, such as 10% for a car indicating a potential threat, and adjusting these percentages based on the speed of travel, the system can determine the level of danger posed by objects on screen.']}, {'end': 404.626, 'start': 310.54, 'title': 'Object detection and classification', 'summary': 'Discusses the process of object detection and classification through identifying classes such as cars, buses, and trucks, and the challenges associated with it, including the difficulty in locating relevant configuration files.', 'duration': 94.086, 'highlights': ['Object detection involves identifying classes such as cars, buses, and trucks, with car being 3, bus being 6, and truck being 8.', "The pedestrian takes up 30% of the screen, and hitting the pedestrian depends on the object's width and type.", 'Challenges in finding relevant configuration files for object detection models.']}], 'duration': 198.659, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA205967.jpg', 'highlights': ['Calculating the exact distance of an object in a 2D image is possible if the size of the object is known, such as in the case of detecting a car with a width between 4 and 5 feet.', 'Using pixel measurements to estimate the size and distance of objects in Grand Theft Auto The method involves measuring the amount of pixels between specific points to determine the relative size and distance of objects, such as cars and pedestrians.', 'Object detection involves identifying classes such as cars, buses, and trucks, with car being 3, bus being 6, and truck being 8.']}, {'end': 671.77, 'segs': [{'end': 452.009, 'src': 'embed', 'start': 428.917, 'weight': 0, 'content': [{'end': 438.502, 'text': "So I just went there and figured out what was the class for car, truck, and bus, but there's 90 possible classifications.", 'start': 428.917, 'duration': 9.585}, {'end': 444.825, 'text': "So there's a lot of different things you could add, but I figured kind of like car, truck, bus, they all have to fit in standard American lanes.", 'start': 438.542, 'duration': 6.283}, {'end': 447.827, 'text': "So, they're probably about the same-ish width.", 'start': 445.345, 'duration': 2.482}, {'end': 452.009, 'text': 'Cars are going to be less width, obviously, than like trucks, especially like a semi-truck or something like that.', 'start': 447.907, 'duration': 4.102}], 'summary': 'Identified 90 possible classifications for car, truck, and bus, based on standard american lane widths.', 'duration': 23.092, 'max_score': 428.917, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA428917.jpg'}, {'end': 521.719, 'src': 'embed', 'start': 495.395, 'weight': 1, 'content': [{'end': 502.225, 'text': 'but actually, if you lower that threshold, you would be astonished by how many objects are actually being detected on your screen.', 'start': 495.395, 'duration': 6.83}, {'end': 503.346, 'text': "It's, it's a large list.", 'start': 502.305, 'duration': 1.041}, {'end': 512.589, 'text': "Um, So you can change that, but anyways, but if we just iterate through these, you'll get tons of objects that have less than 50%,", 'start': 504.228, 'duration': 8.361}, {'end': 514.371, 'text': "and we just don't really want those.", 'start': 512.589, 'duration': 1.782}, {'end': 521.719, 'text': 'You can feel free to tweak this a little more, if you like, especially the further away you get from real,', 'start': 515.292, 'duration': 6.427}], 'summary': 'Lowering the threshold increases object detection, but filtering out objects with less than 50% certainty is necessary.', 'duration': 26.324, 'max_score': 495.395, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA495395.jpg'}, {'end': 558.149, 'src': 'embed', 'start': 532.851, 'weight': 3, 'content': [{'end': 538.194, 'text': "I could also call this relative distance, because we're not actually going to convert this to any metric or imperial units.", 'start': 532.851, 'duration': 5.343}, {'end': 540.456, 'text': "I'm just going to use a relative distance.", 'start': 538.395, 'duration': 2.061}, {'end': 545.42, 'text': "You can feel free to write a conversion if you'd like, something that's fairly realistic.", 'start': 540.856, 'duration': 4.564}, {'end': 546.12, 'text': "I'd love to see it.", 'start': 545.48, 'duration': 0.64}, {'end': 550.423, 'text': 'Because it gets hard as things kind of travel off into the distance.', 'start': 547.441, 'duration': 2.982}, {'end': 558.149, 'text': "And I thought about doing it and I was like, there's really no point because we don't care about how things are like way off in the distance.", 'start': 550.523, 'duration': 7.626}], 'summary': 'Discussion on relative distance without conversion to metric or imperial units.', 'duration': 25.298, 'max_score': 532.851, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA532851.jpg'}], 'start': 411.311, 'title': 'Object detection and measurement', 'summary': 'Discusses configuring class links, 90 possible classifications, object width comparisons, object detection scores, and measuring approximate distance in relative units, emphasizing the importance of setting detection score thresholds and the challenges of measuring distance in relative units.', 'chapters': [{'end': 671.77, 'start': 411.311, 'title': 'Object detection and measurement', 'summary': 'Discusses configuring class links, 90 possible classifications, object width comparisons, object detection scores, and measuring approximate distance in relative units, emphasizing the importance of setting detection score thresholds and the challenges of measuring distance in relative units.', 'duration': 260.459, 'highlights': ['Configuring class links and identifying 90 possible classifications The speaker mentions configuring class links and identifying 90 possible classifications for objects, demonstrating a wide range of potential classifications for objects.', 'Comparing object widths and the standard American lane fit The speaker discusses the comparison of object widths such as cars, trucks, and buses, emphasizing that they generally fit in standard American lanes and noting the differences in width between cars and trucks.', 'Setting object detection score threshold and its impact The chapter emphasizes the significance of setting a minimum score required to draw objects, highlighting the impact of adjusting the score threshold on the number of detected objects and the need to avoid objects with scores less than 50%.', 'Challenges of measuring approximate distance in relative units The speaker discusses the challenges of measuring approximate distance in relative units, highlighting the limitations of using relative distance and the need for realistic conversions as objects move off into the distance.']}], 'duration': 260.459, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA411311.jpg', 'highlights': ['Configuring class links and identifying 90 possible classifications for objects, demonstrating a wide range of potential classifications for objects.', 'Setting object detection score threshold and its impact, emphasizing the significance of setting a minimum score required to draw objects and the impact of adjusting the score threshold on the number of detected objects.', 'Comparing object widths such as cars, trucks, and buses, emphasizing that they generally fit in standard American lanes and noting the differences in width between cars and trucks.', 'Challenges of measuring approximate distance in relative units, highlighting the limitations of using relative distance and the need for realistic conversions as objects move off into the distance.']}, {'end': 902.865, 'segs': [{'end': 734.437, 'src': 'embed', 'start': 671.77, 'weight': 0, 'content': [{'end': 685.113, 'text': "now the other thing we'd like to do is let's go ahead and in case this in parentheses and then we'll do a one, minus that number and then this way,", 'start': 671.77, 'duration': 13.343}, {'end': 689.115, 'text': 'because, like this way would have been a larger as it got closer.', 'start': 685.113, 'duration': 4.002}, {'end': 696.318, 'text': "and but if we do one minus the difference here, um, it'll be smaller as it gets closer.", 'start': 689.115, 'duration': 7.203}, {'end': 698.619, 'text': "so that's just a little more intuitive, if you ask me.", 'start': 696.318, 'duration': 2.301}, {'end': 709.155, 'text': "Okay, so, once we have that, what we're going to go ahead and do.", 'start': 700.788, 'duration': 8.367}, {'end': 709.775, 'text': "I think that's right.", 'start': 709.155, 'duration': 0.62}, {'end': 711.537, 'text': 'Let me just make sure that closes off.', 'start': 710.296, 'duration': 1.241}, {'end': 721.685, 'text': 'Okay, now to give it a little more granularity as we get close, another option we have is to encase it like this.', 'start': 712.638, 'duration': 9.047}, {'end': 728.791, 'text': 'One minus, and then to the power of four, or something like that.', 'start': 724.107, 'duration': 4.684}, {'end': 730.352, 'text': "That'll give us a little more granularity.", 'start': 728.831, 'duration': 1.521}, {'end': 734.437, 'text': 'Cool So good enough.', 'start': 731.874, 'duration': 2.563}], 'summary': 'Discussing mathematical operations and increasing granularity for better accuracy.', 'duration': 62.667, 'max_score': 671.77, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA671770.jpg'}, {'end': 794.216, 'src': 'embed', 'start': 762.621, 'weight': 2, 'content': [{'end': 763.822, 'text': "Where's the middle point of that car?", 'start': 762.621, 'duration': 1.201}, {'end': 766.905, 'text': 'So we can determine is this car actually in our sights?', 'start': 764.202, 'duration': 2.703}, {'end': 769.707, 'text': 'Is it far off to the right or far off to the left??', 'start': 767.005, 'duration': 2.702}, {'end': 772.33, 'text': "Or is it something we're about to hit?", 'start': 769.747, 'duration': 2.583}, {'end': 776.974, 'text': "And if we wanted to avoid it, if it's more to our right, then we could turn left or something like that.", 'start': 772.35, 'duration': 4.624}, {'end': 779.857, 'text': 'So we kind of want to know the midpoint of that vehicle.', 'start': 777.014, 'duration': 2.843}, {'end': 789.006, 'text': "So the way that we're going to do that is I'm going to say mid underscore X is equal to, and in fact, I'm going to just take this here.", 'start': 780.458, 'duration': 8.548}, {'end': 794.216, 'text': "It's equal to the X's, but we're actually just going to add them to each other.", 'start': 790.475, 'duration': 3.741}], 'summary': 'Determining the midpoint of the car to assess its position and potential collision avoidance through x-coordinate calculation.', 'duration': 31.595, 'max_score': 762.621, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA762621.jpg'}], 'start': 671.77, 'title': 'Optimizing calculation precision and determining middle point of an object', 'summary': 'Discusses optimizing calculation precision through a specific formula for granular results and calculating the middle point of a car using x and y coordinates for navigation and collision avoidance, displayed on a screen with specific pixel coordinates.', 'chapters': [{'end': 734.437, 'start': 671.77, 'title': 'Optimizing calculation precision', 'summary': 'Discusses optimizing calculation precision by using a specific formula to adjust values as they get closer, aiming for a more intuitive and granular result.', 'duration': 62.667, 'highlights': ['Using a formula to adjust values as they get closer, making it more intuitive and smaller, which can be achieved by doing one minus the difference (quantifiable data: smaller values as it gets closer)', 'Encasing the calculation in a specific way and using a power of four to achieve a more granular result (quantifiable data: increased granularity)']}, {'end': 902.865, 'start': 734.538, 'title': 'Calculating middle point of object', 'summary': 'Discusses calculating the middle point of a car using x and y coordinates, as well as determining its distance and position for navigation, to avoid potential collisions, and display the information on a screen with specific pixel coordinates.', 'duration': 168.327, 'highlights': ['The chapter discusses calculating the middle point of a car using X and Y coordinates, as well as determining its distance and position for navigation, to avoid potential collisions, and display the information on a screen with specific pixel coordinates.', "The midpoint of the vehicle is calculated by taking the average of its X and Y coordinates, which can be used to determine the car's position and distance for navigation and collision avoidance.", 'The calculated midpoint of the car is utilized to assess its position in relation to the observer, enabling potential navigation adjustments to avoid collisions.']}], 'duration': 231.095, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA671770.jpg', 'highlights': ['Using a formula to adjust values as they get closer, making it more intuitive and smaller, achieved by doing one minus the difference (smaller values as it gets closer)', 'Encasing the calculation in a specific way and using a power of four to achieve a more granular result (increased granularity)', 'The midpoint of the vehicle is calculated by taking the average of its X and Y coordinates, used for navigation and collision avoidance', 'The calculated midpoint of the car is utilized to assess its position in relation to the observer, enabling potential navigation adjustments to avoid collisions']}, {'end': 1156.317, 'segs': [{'end': 938.558, 'src': 'embed', 'start': 903.525, 'weight': 0, 'content': [{'end': 906.927, 'text': "So we're going to say mid x times 800.", 'start': 903.525, 'duration': 3.402}, {'end': 913.392, 'text': 'And then int mid y times 450.', 'start': 906.927, 'duration': 6.465}, {'end': 916.114, 'text': "Just make sure I wasn't off screen.", 'start': 913.392, 'duration': 2.722}, {'end': 919.216, 'text': 'And then we can pick the font.', 'start': 918.035, 'duration': 1.181}, {'end': 922.218, 'text': "So we'll just do cv2.allcaps.", 'start': 919.376, 'duration': 2.842}, {'end': 925.14, 'text': 'Font underscore Hershey underscore simplex.', 'start': 922.278, 'duration': 2.862}, {'end': 930.934, 'text': "And then what we're going to do is for the size, we'll do 0.7.", 'start': 926.431, 'duration': 4.503}, {'end': 933.314, 'text': "The color, we'll do 255, 255, 255.", 'start': 930.934, 'duration': 2.38}, {'end': 938.558, 'text': 'And then line width will be a 2.', 'start': 933.315, 'duration': 5.243}], 'summary': 'Using cv2.allcaps font with size 0.7 and color 255, 255, 255.', 'duration': 35.033, 'max_score': 903.525, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA903525.jpg'}, {'end': 993.436, 'src': 'embed', 'start': 961.21, 'weight': 1, 'content': [{'end': 973.977, 'text': "the final thing we're going to say is, if apx distance is less than or equal to um,", 'start': 961.21, 'duration': 12.767}, {'end': 977.399, 'text': "we'll just start with 0.5 and then we'll kind of see how we're doing so.", 'start': 973.977, 'duration': 3.422}, {'end': 980.098, 'text': "it's less than or equal to 0.5.", 'start': 977.399, 'duration': 2.699}, {'end': 985.826, 'text': "um, that object is close, but that object, if it's like off to our side,", 'start': 980.098, 'duration': 5.728}, {'end': 993.436, 'text': "like a little bit like if it's like near our like front quarter panel of the car, um, it's going to be huge, but we're missing it.", 'start': 985.826, 'duration': 7.61}], 'summary': "If apx distance ≤ 0.5, we'll start with 0.5. object close if near front quarter panel of the car.", 'duration': 32.226, 'max_score': 961.21, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA961210.jpg'}, {'end': 1156.317, 'src': 'heatmap', 'start': 1113.584, 'weight': 2, 'content': [{'end': 1118.827, 'text': "All right, let's see if our 70% GPU usage works better.", 'start': 1113.584, 'duration': 5.243}, {'end': 1121.568, 'text': "So we're in a car, we're detecting cars at distance.", 'start': 1119.327, 'duration': 2.241}, {'end': 1124.169, 'text': "That's pretty cool.", 'start': 1123.489, 'duration': 0.68}, {'end': 1129.492, 'text': "As it was passing on our side, it wasn't a problem, but then it became a problem.", 'start': 1124.269, 'duration': 5.223}, {'end': 1131.013, 'text': 'No warning for that guy as we pass.', 'start': 1129.572, 'duration': 1.441}, {'end': 1132.494, 'text': "Let's go ahead and head towards these cars.", 'start': 1131.033, 'duration': 1.461}, {'end': 1134.655, 'text': 'We got a warning.', 'start': 1134.114, 'duration': 0.541}, {'end': 1136.776, 'text': 'It was probably a little too late, but it worked.', 'start': 1134.695, 'duration': 2.081}, {'end': 1140.898, 'text': 'My minus 50 actually was a pretty good distance there.', 'start': 1137.776, 'duration': 3.122}, {'end': 1144.487, 'text': 'those warnings are way too late.', 'start': 1143.346, 'duration': 1.141}, {'end': 1149.451, 'text': "i just tried to break, like as soon as i saw the warning, it didn't work.", 'start': 1144.487, 'duration': 4.964}, {'end': 1156.317, 'text': "but what we can do is, just since that guy's no longer with us, so we can use this car for an example, which is apparently a truck, but that's okay.", 'start': 1149.451, 'duration': 6.866}], 'summary': 'Testing 70% gpu usage for car detection, with warnings being late and not effective.', 'duration': 46.699, 'max_score': 1113.584, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA1113584.jpg'}], 'start': 903.525, 'title': 'Opencv, object proximity, and car detection', 'summary': "Discusses overlaying text on images using opencv, displaying warnings based on object proximity, and testing an autonomous car's ability to detect cars at a distance for debugging and safety purposes.", 'chapters': [{'end': 961.21, 'start': 903.525, 'title': 'Opencv text overlay', 'summary': 'Discusses the process of overlaying text on an image using opencv, specifying the font, size, color, and line width, to display the approximate relative distance on a car for debugging purposes.', 'duration': 57.685, 'highlights': ['The process of overlaying text on an image using OpenCV involves specifying the font, size, color, and line width, with the goal of displaying the approximate relative distance on a car for debugging purposes.', "The font specified for the text overlay is 'Hershey Simplex' with a size of 0.7 and a color of 255, 255, 255.", 'The line width for the text overlay is set to 2.']}, {'end': 1109.558, 'start': 961.21, 'title': 'Display warning based on object proximity', 'summary': 'Discusses the process of displaying a warning message based on the proximity of an object to the vehicle, utilizing distance measurements and object positioning to determine when to display the warning.', 'duration': 148.348, 'highlights': ['The system checks if the distance from the object is less than or equal to 0.5 for determining proximity.', "A conditional check is made based on the position of the object relative to the vehicle's front quarter panel, with a range of 0.3 to 0.7 for the mid X value.", 'The warning message is displayed using CV2 with specific color and position parameters.']}, {'end': 1156.317, 'start': 1109.618, 'title': 'Autonomous car test: detecting cars at distance', 'summary': "Discusses testing an autonomous car's ability to detect cars at a distance, with a focus on warning system effectiveness and the success rate of braking in response to warnings.", 'duration': 46.699, 'highlights': ["The warning system was late in alerting the driver, with warnings being deemed 'way too late.'", 'The braking system did not work as intended when responding to the warning, resulting in the need for earlier warnings or improved braking response.', "Success was found in detecting cars at a distance, with the warning system proving effective despite being 'probably a little too late.'", "The 70% GPU usage was tested to see if it improved the car's performance during the detection of cars at distance."]}], 'duration': 252.792, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA903525.jpg', 'highlights': ['The process of overlaying text on an image using OpenCV involves specifying the font, size, color, and line width, with the goal of displaying the approximate relative distance on a car for debugging purposes.', 'The system checks if the distance from the object is less than or equal to 0.5 for determining proximity.', "Success was found in detecting cars at a distance, with the warning system proving effective despite being 'probably a little too late.'", "The font specified for the text overlay is 'Hershey Simplex' with a size of 0.7 and a color of 255, 255, 255.", "The warning system was late in alerting the driver, with warnings being deemed 'way too late.'", "The 70% GPU usage was tested to see if it improved the car's performance during the detection of cars at distance."]}, {'end': 1278.31, 'segs': [{'end': 1230.177, 'src': 'embed', 'start': 1178.246, 'weight': 0, 'content': [{'end': 1187.299, 'text': "now it's this car, and if we continue to turn, i bet both of these would probably could be a warning, maybe sort of If I zoom back, there we go.", 'start': 1178.246, 'duration': 9.053}, {'end': 1188.821, 'text': 'Now both cars are a warning.', 'start': 1187.58, 'duration': 1.241}, {'end': 1197.689, 'text': 'So pretty cool that we can detect other cars, detect kind of some sort of, have some sort of collision warning system and all that.', 'start': 1189.261, 'duration': 8.428}, {'end': 1200.932, 'text': 'It also looks like the frame rates are better.', 'start': 1198.951, 'duration': 1.981}, {'end': 1204.916, 'text': 'That works.', 'start': 1204.216, 'duration': 0.7}, {'end': 1210.482, 'text': "So 70% seems to be good enough to record it, so I'll just continue doing that so I don't kill my GPU.", 'start': 1205.056, 'duration': 5.426}, {'end': 1213.27, 'text': 'OK, pretty cool.', 'start': 1212.21, 'duration': 1.06}, {'end': 1218.533, 'text': "But as I was saying before, that's not all we can do with the Object Detection API.", 'start': 1214.791, 'duration': 3.742}, {'end': 1223.675, 'text': "There's obviously a lot of stuff that we can do within the kind of self-driving car format.", 'start': 1218.653, 'duration': 5.022}, {'end': 1230.177, 'text': "But one of the other things I'd really like to do is at least for the stream.", 'start': 1224.855, 'duration': 5.322}], 'summary': 'Object detection api can detect cars with 70% accuracy, improving frame rates, and avoiding gpu overload.', 'duration': 51.931, 'max_score': 1178.246, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA1178246.jpg'}, {'end': 1274.429, 'src': 'embed', 'start': 1251.385, 'weight': 2, 'content': [{'end': 1258.546, 'text': "Could we write some code that accurately goes to the car and steals the car if we're on foot?", 'start': 1251.385, 'duration': 7.161}, {'end': 1262.207, 'text': "So that's what I'm going to do in the next probably two tutorials.", 'start': 1259.106, 'duration': 3.101}, {'end': 1264.167, 'text': 'So you can stay tuned for that.', 'start': 1262.327, 'duration': 1.84}, {'end': 1269.848, 'text': "If you've got questions, comments, concerns, whatever up to the point of this tutorial, let me know.", 'start': 1264.287, 'duration': 5.561}, {'end': 1271.109, 'text': "I'll do my best to help you out.", 'start': 1269.868, 'duration': 1.241}, {'end': 1274.429, 'text': 'Otherwise, I will see you in the next tutorial.', 'start': 1271.689, 'duration': 2.74}], 'summary': 'Planning to write code to steal a car in next two tutorials.', 'duration': 23.044, 'max_score': 1251.385, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA1251385.jpg'}], 'start': 1156.317, 'title': 'Object detection api and self-driving car', 'summary': 'Discusses the detection of cars, improvement in frame rates, and potential system development for stealing cars using the object detection api, with plans for future tutorials on the topic.', 'chapters': [{'end': 1278.31, 'start': 1156.317, 'title': 'Object detection api and self-driving car', 'summary': 'Discusses the detection of cars, the improvement in frame rates, and the potential to develop a system for stealing cars using the object detection api, with plans for future tutorials on the topic.', 'duration': 121.993, 'highlights': ['The Object Detection API can detect other cars and has a collision warning system. The API is capable of detecting other cars and providing a collision warning system, showcasing its functionality in a real-time scenario.', 'Improvement in frame rates to 70% for better recording without overloading the GPU. The frame rates have improved to 70%, allowing for better recording while preventing GPU overload, enhancing performance and efficiency.', 'Future plan to develop a system using the Object Detection API to steal cars if on foot. The plan includes using the Object Detection API to develop a system for stealing cars when on foot, indicating a unique and innovative application of the technology.']}], 'duration': 121.993, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/o3Ky_EdHVrA/pics/o3Ky_EdHVrA1156317.jpg', 'highlights': ['The Object Detection API can detect other cars and has a collision warning system.', 'Improvement in frame rates to 70% for better recording without overloading the GPU.', 'Future plan to develop a system using the Object Detection API to steal cars if on foot.']}], 'highlights': ['The video aims to enhance object detection in Grand Theft Auto by detecting cars in the distance and calculating their proximity to take necessary actions.', 'Using a formula to adjust values as they get closer, making it more intuitive and smaller, achieved by doing one minus the difference (smaller values as it gets closer)', 'The midpoint of the vehicle is calculated by taking the average of its X and Y coordinates, used for navigation and collision avoidance', 'Configuring class links and identifying 90 possible classifications for objects, demonstrating a wide range of potential classifications for objects.', 'The Object Detection API can detect other cars and has a collision warning system.']}