Update README.md (0d9cd5f9) · Commits · IntelligentTEAM / AAAI2021 Locount Dataset

README.md

+5 −5

Original line number	Diff line number	Diff line
		@@ -28,7 +28,7 @@ To solve the above issues, we collect a large-scale object localization and coun
		More than 1.9 million object instances in 140 categories (including Jacket, Shoes, Oven, etc.) are annotated.


		<div align=center><img src="Images/dataset-summary.png" width="1200" height="400"/></div>
		<div align=center><img src="Images/dataset-summary.png" width="1000" height="400"/></div>
		Table 1: Summary of existing object detection benchmarks in retail stores. “C” indicates the image classification task, “S”
		indicates the single-class object detection task, and “M” indicates the multi-class object detection task.

		@@ -41,7 +41,7 @@ There are various factors challenging the performance of algorithms, such as sca



		<div align=center><img src="Images/category-tree.png" width="800" height="400"/></div>
		<div align=center><img src="Images/category-tree.png" width="900" height="400"/></div>
		<p align="center">Figure 2: Category hierarchy of the large-scale localization and counting dataset in
		the shelf scenarios.</p>

		@@ -62,7 +62,7 @@ multiple stages for localization and counting, i.e., S_{1},..., S_{N} are cascad
		instances enclosed in the bounding box, where N is the total number of stages. For more detailed definitions, please refer to the [paper](http://arxiv.org/abs/2003.08230).
		The counting accuracy threshold for the positive/negative sample generation is determined by the architecture design of CLCNet, which is described as follows.

		<div align=center><img src="Images/framework.png" width="700" height="400" /></div>
		<div align=center><img src="Images/framework.png" width="800" height="400" /></div>
		<p align="center">Figure 3: The architecture of our CLCNet for the Locount task. The cubes indicate the output feature maps from the convolutional layers or RoIAlign operation.
		The numbers in the brackets indicate the range of counting number in each stage.</p>

		@@ -79,11 +79,11 @@ Thus, the counting task is formulated as the multi-class classification task, wh

		We conduct several experiments of the state-of-the-art object detectors and the proposed CLCNet method on the proposed dataset, to demonstrate the effectiveness of CLCNet, Table 2 and Fig. 4.

		<div align=center><img src="Images/Experiment-results.png" width="700" height="400" /></div>
		<div align=center><img src="Images/Experiment-results.png" width="800" height="400" /></div>
		Table 2: Comparison results of the algorithms on the proposed dataset. Detection results of all comparison methods on the
		proposed dataset. The mark lc on the upper right corner indicates that its value is computed by the proposed metrics

		<div align=center><img src="Images/show-results.jpg" width="700" height="400" /></div>
		<div align=center><img src="Images/show-results.jpg" width="800" height="400" /></div>
		Figure 4: Qualitative results of the proposed CLCNet method on the Locount dataset.