What is this?
A video-based deep learning course that walks through classic computer-vision architectures for image classification, object detection, segmentation and keypoint detection, pairing each network with a lecture and runnable code.
A video-based deep learning course that walks through classic computer-vision architectures for image classification, object detection, segmentation and keypoint detection, pairing each network with a lecture and runnable code.Is this a ready-made library, or a set of teaching examples?It is a teaching resource rather than an installable package: each network has its own PyTorch implementation and, for most of them, a separate TensorFlow2 (keras) implementation to study side by side.
It is a teaching resource rather than an installable package: each network has its own PyTorch implementation and, for most of them, a separate TensorFlow2 (keras) implementation to study side by side.How do I follow along with a lesson?Pick a network from the course list (e.g. ResNet, YOLO, U-Net), watch the matching bilibili video explanation, then open the corresponding pytorch_* or tensorflow_* folder in the repo to run the code yourself; lecture slides are kept in the courseppt folder.
Pick a network from the course list (e.g. ResNet, YOLO, U-Net), watch the matching bilibili video explanation, then open the corresponding pytorch_* or tensorflow_* folder in the repo to run the code yourself; lecture slides are kept in the courseppt folder.Do I need both PyTorch and TensorFlow2 to use this?No, you only need the framework you want to study — most networks are implemented in both PyTorch and TensorFlow2, so you can stick to one of them.
No, you only need the framework you want to study — most networks are implemented in both PyTorch and TensorFlow2, so you can stick to one of them.Are the video lectures in English?No, the course, its slides and the videos are all in Chinese and hosted on the author's bilibili channel.
No, the course, its slides and the videos are all in Chinese and hosted on the author's bilibili channel.Who is this course meant for?Students and self-learners who want to understand and reimplement classic CV architectures from scratch, since the material comes from the author's own graduate-school research notes.
Students and self-learners who want to understand and reimplement classic CV architectures from scratch, since the material comes from the author's own graduate-school research notes.What topics does the course cover overall?Four areas: image classification (LeNet through ConvNeXt and MobileViT), object detection (Faster-RCNN/FPN, SSD, RetinaNet, the YOLO series, FCOS), semantic segmentation (FCN, DeepLabV3, U-Net, U2Net) and instance/keypoint tasks (Mask R-CNN, DeepPose, HRNet).
Four areas: image classification (LeNet through ConvNeXt and MobileViT), object detection (Faster-RCNN/FPN, SSD, RetinaNet, the YOLO series, FCOS), semantic segmentation (FCN, DeepLabV3, U-Net, U2Net) and instance/keypoint tasks (Mask R-CNN, DeepPose, HRNet).Where can I find the slides used in the lectures?All PPT slides are stored in the courseppt folder of the repository and can be downloaded directly.
All PPT slides are stored in the courseppt folder of the repository and can be downloaded directly.Is the course still growing?Yes, entries in the course outline are marked as completed once their video and code are finished, and the author notes new topics will be added as they are studied.
Yes, entries in the course outline are marked as completed once their video and code are finished, and the author notes new topics will be added as they are studied.What are similar projects?mmdetection, Swin-Transformer, DeepLearning, vit-pytorch, X-AnyLabeling, VisionHOPE, awesome-deep-vision, deeplearning_ai_books
mmdetection, Swin-Transformer, DeepLearning, vit-pytorch, X-AnyLabeling, VisionHOPE, awesome-deep-vision, deeplearning_ai_booksWhich topics is it in?swin-transformer, mask-rcnn, tensorflow2, instance-segmentation, vision-transformer, convolutional-neural-networks, semantic-segmentation, yolo, cnn, deep-neural-networks, bilibili, segmentation
swin-transformer, mask-rcnn, tensorflow2, instance-segmentation, vision-transformer, convolutional-neural-networks, semantic-segmentation, yolo, cnn, deep-neural-networks, bilibili, segmentation