Thanks Eric. Are you running YOLO on the rpi5? I’m waiting for Arducam to combine their tof depth camera with an rgb camera in a single module. You don’t really need a stereo camera if you have a tof. Also, hiwonder has a couple of cheaper (than oak) depth camera on their website. I’m keen to try one.
Hi. Unfortunately, since it is not equipped with a stereo-pair camera, we cannot obtain information about depth estimation. Have you heard that Arducam plans to release a camera with a ToF module?
@@Alex_Anderbrücke PiNSIGHT is based on Luxonis OAK-SoM and helps to easily configure NN through their module. Pi Ai Camera is also based on RP2040 and helps to configure NN. The configuration method is the same, but the program used to create AI models is different. I think there will be differences in application development due to this. I will try it if I have a chance. Thank you.
This is an AI camera for Raspberry Pi, and a microcontroller is built-in inside the camera, making it easy to implement and use AI functions. If you do not use AI-related features, it may be better to use a regular USB camera. www.arducam.com/product/presalesarducam-pinsight-12mp-vision-ai-mate-for-raspberry-pi-5/
@@Gerald-iz7mv Of course, it is possible to acquire image frames directly through OpenCV. You can use various frameworks with this camera, including pytorch and TensorFlow. docs.luxonis.com/projects/api/en/latest/samples/host_side/opencv_support/
@@Gerald-iz7mv I can't answer that for sure because there may be better cameras out there that I don't know about. Anyway, this is definitely a great camera that runs on Raspberry Pi 4/5. You can find more products here. www.arducam.com/usb-board-cameras-uvc-modules-webcams/
Hi Eric, I would like to use an IP camera + RP5 to recognize objects fast and accurate. Is it correct that the RP5 + IP-cam only can not do the job well? Do I need a kind of GPU? Someone mentioned Yolo, is that the best option? Or would you recommend another setup if an IPcam is used? In my project I need to see if there is a bike seen by the camera or not, any suggestions? Could you do a video about that please? Thanks.
If you need to prepare a model for Yolo and train it directly on your system, you will need a GPU, but it doesn't seem to be necessary for your project. My question is why would you want to use IPCam? Does the camera need to connect wirelessly?
@@ThatProject The ip-cam is strong , high quality easy to mount and weather proof. Also the idea to be able to apply the solution to every existing ip cam is beneficial. yolo has a pre-trained model. A reaction time of 1 sec is ok, just wonder if gpu is needed and will only affect reaction time or also the accuracy?
@@SA-oj3bo From what I've seen, if your project uses pre-trained models and simply detects specific objects, the GPU won't play much of a role. If you need to find many objects in an image and require accuracy, an environment with a GPU is advantageous.
![BOWKYLION - วิงวอน (ex-change) [Official MV]](http://i.ytimg.com/vi/cLdnZWDaO-w/mqdefault.jpg)
Thanks Eric. Are you running YOLO on the rpi5? I’m waiting for Arducam to combine their tof depth camera with an rgb camera in a single module. You don’t really need a stereo camera if you have a tof. Also, hiwonder has a couple of cheaper (than oak) depth camera on their website. I’m keen to try one.
Hi. Unfortunately, since it is not equipped with a stereo-pair camera, we cannot obtain information about depth estimation. Have you heard that Arducam plans to release a camera with a ToF module?
@@ThatProject I haven’t heard anything about it. Just wishful thinking,
Would you mind sharing the info on the HDMI display used in the video?
It's MPI3508.
could be very nice of you to create step by step guide to to setup this camera rather that short overview
Agree. This is just an introduction to a new product. Once the fun project is completed, I will make a tutorial about it. Thanks.
@@ThatProject sounds good bro, will be waiting
kindly compare this setting with a new Raspberry pi AI camera, which just came out from Sony/Raspberry Pi official. what would be the difference?
@@Alex_Anderbrücke PiNSIGHT is based on Luxonis OAK-SoM and helps to easily configure NN through their module. Pi Ai Camera is also based on RP2040 and helps to configure NN. The configuration method is the same, but the program used to create AI models is different. I think there will be differences in application development due to this. I will try it if I have a chance. Thank you.
This or nano jetson will be better for AI?
@@dawidpodziewski4157 Nano Jetson is much more powerful.
Is that the rasbperry pi ai camera? Where you ordered it? Can you use the camera also without ai?
This is an AI camera for Raspberry Pi, and a microcontroller is built-in inside the camera, making it easy to implement and use AI functions. If you do not use AI-related features, it may be better to use a regular USB camera. www.arducam.com/product/presalesarducam-pinsight-12mp-vision-ai-mate-for-raspberry-pi-5/
@@ThatProject if i use the ai camera i can use it with ai and without ai features? Lets say i get the raw image and do some opencv with ai myself…
@@Gerald-iz7mv Of course, it is possible to acquire image frames directly through OpenCV. You can use various frameworks with this camera, including pytorch and TensorFlow. docs.luxonis.com/projects/api/en/latest/samples/host_side/opencv_support/
@@ThatProjecti dont see a better camera than this new ai camera for rpi which is faster - do you? Have you looked into thermal cameras for the rpi?
@@Gerald-iz7mv I can't answer that for sure because there may be better cameras out there that I don't know about. Anyway, this is definitely a great camera that runs on Raspberry Pi 4/5. You can find more products here. www.arducam.com/usb-board-cameras-uvc-modules-webcams/
Hi Eric, I would like to use an IP camera + RP5 to recognize objects fast and accurate. Is it correct that the RP5 + IP-cam only can not do the job well? Do I need a kind of GPU? Someone mentioned Yolo, is that the best option? Or would you recommend another setup if an IPcam is used? In my project I need to see if there is a bike seen by the camera or not, any suggestions? Could you do a video about that please? Thanks.
If you need to prepare a model for Yolo and train it directly on your system, you will need a GPU, but it doesn't seem to be necessary for your project. My question is why would you want to use IPCam? Does the camera need to connect wirelessly?
@@ThatProject The ip-cam is strong , high quality easy to mount and weather proof. Also the idea to be able to apply the solution to every existing ip cam is beneficial. yolo has a pre-trained model. A reaction time of 1 sec is ok, just wonder if gpu is needed and will only affect reaction time or also the accuracy?
@@SA-oj3bo From what I've seen, if your project uses pre-trained models and simply detects specific objects, the GPU won't play much of a role. If you need to find many objects in an image and require accuracy, an environment with a GPU is advantageous.